WO2025040048A1 - Data processing method and apparatus - Google Patents

Abstract

A data processing method, relating to the field of artificial intelligence, and comprising: acquiring an image and an object detection request, the object detection request containing a natural language description of an object to be detected in the image; by means of an image encoder, processing the image to obtain a feature representation of each image region among a plurality of image regions, each image region corresponding to a candidate detection box; and, by means of a language model, processing the object detection request and the multiple feature representations to determine from among the plurality of image regions a region where said object is located and the category of said object. The present application uses the image encoder to obtain fine-grained features, i.e. the features of each image region, and performs object detection on the basis of the language model, thereby improving the capability of processing the fine-grained features and the precision of object detection under the guidance of human languages.

Description

A data processing method and device thereof

This application claims the priority of the Chinese patent application filed with the State Intellectual Property Office on August 23, 2023, with application number 202311071646.4 and application name “A data processing method and device thereof”, the entire contents of which are incorporated by reference in this application.

Technical Field

The present application relates to the field of artificial intelligence, and in particular to a data processing method and device thereof.

Background Art

In the current field of machine vision, open domain detection models, such as DetCLIP, have demonstrated powerful capabilities and can achieve accurate target detection for any given category. However, this model is not impeccable, and there are some obvious problems in its practical application. First, we often cannot provide some specific object categories. This is because there are many kinds of objects in the real world, but our knowledge base and vocabulary are limited. Secondly, the category tables we list are often incomplete. This problem is more due to the limitations of our knowledge acquisition and organization methods. Finally, we cannot give specific categories based on some attributes, functions and other knowledge of the category. This is because the current machine vision system lacks the ability to deeply understand and comprehensively apply knowledge.

However, there are already some large language models, such as ChatGPT, and multimodal large models, such as miniGPT4, which have excellent knowledge understanding and reasoning capabilities. These models can understand and process human natural language, and can also make meaningful inferences based on given information. But they also have some shortcomings, such as the lack of predictive capabilities for fine-grained tasks such as object detection.

Summary of the invention

The present application provides a data processing method that can improve the processing capability for fine-grained features and improve the accuracy of target detection under the guidance of human language.

In a first aspect, the present application provides a data processing method, comprising: acquiring an image and an object detection request, the object detection request comprising a natural language description of an object to be detected in the image; processing the image through an image encoder to obtain a feature representation of each of a plurality of image regions, each image region corresponding to a candidate detection box; processing the object detection request and a plurality of the feature representations through a language model, and determining the region and category of the object to be detected from the plurality of image regions.

In the existing implementation, when performing target detection, the image encoder outputs the feature representation of the entire image, and the related training method makes the language model not have/not good at fine-grained image processing capabilities. In the embodiment of the present application, the image encoder outputs the feature representation of each of the multiple image areas in the image, and each image area can correspond to a candidate detection box, and each detection box can contain an object. The area where the object to be detected indicated in the object detection request is located can be selected from multiple image areas. Specifically, the feature representation of each image area and the object detection request can be input into the language model, and the language model determines whether the object included in each image area is the object to be identified indicated by the object detection request. The language model can also output the category of the object included in the image area when it is determined that the object included in the image area is the object to be identified indicated by the object detection request.

Using the image encoder to obtain the features of each image area and combining it with the language model for target detection can improve the processing capabilities for fine-grained features and improve the accuracy of target detection under the guidance of human language.

In a possible implementation, the processing of the object detection request and the multiple feature representations through a language model to determine the area where the object to be detected is located from the multiple image areas includes: processing the object detection request and each feature representation in parallel through a language model to obtain a detection result for each image area; and determining, based on the detection results, the area where the object to be detected is located from the multiple image areas.

In a possible implementation, before the image is processed by the image encoder, the method further includes: obtaining a training sample, the training sample including an image sample and a corresponding text sample, the text sample being a text description corresponding to the image sample; processing the image sample by the image encoder to obtain a first processing result; processing the text sample by the text encoder to obtain a second processing result; and updating the image encoder and the text encoder according to the first processing result and the second processing result through comparative learning. Encoder.

In a possible implementation, the first processing result includes multiple detection boxes and a category of each detection box; the method also includes: updating the image encoder and the text encoder according to the first processing result and the corresponding true value.

In a possible implementation, the language model includes multiple network layers, at least one of the multiple network layers includes a cross-attention layer; the cross-attention layer is used to perform attention interaction between image features and text features.

In a second aspect, the present application provides a data processing device, the device comprising:

An acquisition module, configured to acquire an image and an object detection request, wherein the object detection request includes a natural language description of an object to be detected in the image;

A processing module, configured to process the image through an image encoder to obtain a feature representation of each of a plurality of image regions, each image region corresponding to a candidate detection frame;

The object detection request and the plurality of feature representations are processed by a language model, and a region where the object to be detected is located is determined from among the plurality of image regions.

In a possible implementation, the processing module is specifically configured to:

Processing the object detection request and each of the feature representations in parallel through a language model to obtain a detection result for each of the image regions;

According to the detection result, the area where the object to be detected is located is determined from the multiple image areas.

In a possible implementation, before the image is processed by the image encoder, the acquisition module is further used to:

Acquire a training sample, wherein the training sample includes an image sample and a corresponding text sample, wherein the text sample is a text description corresponding to the image sample;

The processing module is further used to: process the image sample through an image encoder to obtain a first processing result;

Processing the text sample through a text encoder to obtain a second processing result;

According to the first processing result and the second processing result, the image encoder and the text encoder are updated through comparative learning.

In a possible implementation, the first processing result includes a plurality of detection frames and a category of each detection frame; and the processing module is further configured to:

The image encoder and the text encoder are updated according to the first processing result and the corresponding true value.

In a possible implementation, the language model includes multiple network layers, at least one of the multiple network layers includes a cross-attention layer; the cross-attention layer is used to perform attention interaction between image features and text features.

In a third aspect, an embodiment of the present application provides a data processing device, which may include a memory, a processor, and a bus system, wherein the memory is used to store programs, and the processor is used to execute the programs in the memory to perform the first aspect and any optional method thereof.

In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer-readable storage medium is run on a computer, the computer executes the above-mentioned first aspect and any optional method thereof.

In a fifth aspect, an embodiment of the present application provides a computer program, which, when executed on a computer, enables the computer to execute the above-mentioned first aspect and any optional method thereof.

In a sixth aspect, the present application provides a chip system, which includes a processor for supporting a data processing device to implement the functions involved in the above aspects, such as sending or processing the data involved in the above methods; or information. In one possible design, the chip system also includes a memory, which is used to store program instructions and data necessary for executing the device or training the device. The chip system can be composed of chips, or it can include chips and other discrete devices.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1A is a schematic diagram of a structure of an artificial intelligence main framework;

1B to 1D are schematic diagrams of the application framework of the present application;

FIG2 is a schematic diagram of the application framework of the present application;

FIG3 is a schematic diagram of the application framework of the present application;

FIG4 is a schematic diagram of the application framework of the present application;

FIG5A and FIG5B are schematic diagrams of a network structure of the present application;

FIG6 is a flowchart of a data processing method provided in an embodiment of the present application;

FIG7A is a schematic diagram of text in a training sample provided in an embodiment of the present application;

FIG7B is a schematic diagram of a training process provided in an embodiment of the present application;

FIG8 is a schematic diagram of a structure of a data processing device provided in an embodiment of the present application;

FIG9 is a schematic diagram of a structure of an execution device provided in an embodiment of the present application;

FIG10 is a schematic diagram of a structure of a training device provided in an embodiment of the present application;

FIG. 11 is a schematic diagram of the structure of a chip provided in an embodiment of the present application.

DETAILED DESCRIPTION

The following describes the embodiments of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. The terms used in the implementation mode of the present invention are only used to explain the specific embodiments of the present invention, and are not intended to limit the present invention.

The embodiments of the present application are described below in conjunction with the accompanying drawings. It is known to those skilled in the art that with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, which is only to describe the distinction mode adopted by the objects of the same attributes when describing in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

First, the overall workflow of the artificial intelligence system is described. Please refer to Figure 1A. Figure 1A shows a structural diagram of the main framework of artificial intelligence. The following is an explanation of the above artificial intelligence theme framework from the two dimensions of "intelligent information chain" (horizontal axis) and "IT value chain" (vertical axis). Among them, the "intelligent information chain" reflects a series of processes from data acquisition to processing. For example, it can be a general process of intelligent information perception, intelligent information representation and formation, intelligent reasoning, intelligent decision-making, intelligent execution and output. In this process, the data has undergone a condensation process of "data-information-knowledge-wisdom". The "IT value chain" reflects the value that artificial intelligence brings to the information technology industry from the underlying infrastructure of human intelligence, information (providing and processing technology implementation) to the industrial ecology process of the system.

(1) Infrastructure

The infrastructure provides computing power support for the artificial intelligence system, enables communication with the outside world, and supports it through the basic platform. It communicates with the outside world through sensors; computing power is provided by smart chips (CPU, NPU, GPU, ASIC, FPGA and other hardware acceleration chips); the basic platform includes distributed computing frameworks and networks and other related platform guarantees and support, which can include cloud storage and computing, interconnected networks, etc. For example, sensors communicate with the outside world to obtain data, and these data are provided to the smart chips in the distributed computing system provided by the basic platform for calculation.

(2) Data

The data on the upper layer of the infrastructure is used to represent the data sources in the field of artificial intelligence. The data involves graphics, images, voice, text, and IoT data of traditional devices, including business data of existing systems and perception data such as force, displacement, liquid level, temperature, and humidity.

(3) Data processing

Data processing usually includes data training, machine learning, deep learning, search, reasoning, decision-making and other methods.

Among them, machine learning and deep learning can symbolize and formalize data for intelligent information modeling, extraction, preprocessing, and training.

Reasoning refers to the process of simulating human intelligent reasoning in computers or intelligent systems, using formalized information to perform machine thinking and solve problems based on reasoning control strategies. Typical functions are search and matching.

Decision-making refers to the process of making decisions after intelligent information is reasoned, usually providing functions such as classification, sorting, and prediction.

(4) General capabilities

After the data has undergone the data processing mentioned above, some general capabilities can be further formed based on the results of the data processing, such as an algorithm or a general system, for example, translation, text analysis, computer vision processing, speech recognition, image recognition, etc.

(5) Smart products and industry applications

Smart products and industry applications refer to the products and applications of artificial intelligence systems in various fields. They are the encapsulation of the overall artificial intelligence solution, which productizes intelligent information decision-making and realizes practical applications. Its application areas mainly include: smart terminals, smart transportation, smart medical care, autonomous driving, smart cities, etc.

The present application can be applied to the field of image processing in the field of artificial intelligence. Taking image processing as an example, multiple application scenarios of products will be introduced below.

First, the application scenario of this application is introduced.

This application can be applied to, but is not limited to, applications with image processing functions (hereinafter referred to as image processing applications) or cloud services provided by cloud servers, etc., which are introduced below:

1. Image processing applications

The product form of the embodiment of the present application can be an image processing application, in particular, an application with an image segmentation function. The image processing application can be run on a terminal device or a server on the cloud side.

In one possible implementation, an image processing application can implement tasks such as target detection based on input images and text to obtain a processing result. The text can specify a description of the detected object, and the processing result can be a detection result (for example, including a detection box and a category).

In one possible implementation, a user can open an image processing application installed on a terminal device and input images and text. The image processing application can process the images and text using the method provided in an embodiment of the present application and present the processing results to the user (the presentation method may be but is not limited to display, saving, uploading to the cloud, etc.).

In one possible implementation, a user can open an image processing application installed on a terminal device and input images and text. The image processing application can send the image and text to a server on the cloud side. The server on the cloud side processes the image and text using the method provided in an embodiment of the present application and transmits the processing results back to the terminal device. The terminal device can present the processing results to the user (the presentation method may be but is not limited to display, saving, uploading to the cloud side, etc.).

Next, the image processing application in the embodiment of the present application is introduced from the functional architecture and the product architecture that realizes the functions.

Referring to FIG. 1B , FIG. 1B is a schematic diagram of the functional architecture of an image processing application in an embodiment of the present application:

In a possible implementation, as shown in FIG1B , an image processing application 102 may receive input parameters 101 (e.g., including images and text) and generate processing results 103. The image processing application 102 may be executed on (for example) at least one computer system and includes computer codes, which, when executed by one or more computers, cause the computers to execute the method provided in the embodiments of the present application.

Referring to FIG. 1C , FIG. 1C is a schematic diagram of the physical architecture for running an image processing application in an embodiment of the present application:

Referring to FIG. 1C , FIG. 1C shows a schematic diagram of a system architecture. The system may include a terminal 100 and a server 200. The server 200 may include one or more servers (FIG. 1C is illustrated by taking one server as an example), and the server 200 may provide the method provided in the embodiment of the present application for one or more terminals.

Among them, an image processing application can be installed on the terminal 100. The above application and web page can provide an interface. The terminal 100 can receive relevant parameters entered by the user on the image processing interface and send the above parameters to the server 200. The server 200 can obtain the processing result based on the received parameters and return the processing result to the terminal 100.

It should be understood that in some optional implementations, the terminal 100 can also complete the action of obtaining the processing result based on the received parameters by itself without the cooperation of the server, and the embodiments of the present application are not limited to this.

Next, the product form of the terminal 100 in FIG. 1C is described;

The terminal 100 in the embodiment of the present application can be a mobile phone, a tablet computer, a wearable device, a vehicle-mounted device, an augmented reality (AR)/virtual reality (VR) device, a laptop computer, an ultra-mobile personal computer (UMPC), a netbook, a personal digital assistant (PDA), etc., and the embodiment of the present application does not impose any limitation on this.

FIG. 1D shows a schematic diagram of an optional hardware structure of the terminal 100 .

1D , the terminal 100 may include components such as a radio frequency unit 110, a memory 120, an input unit 130, a display unit 140, a camera 150 (optional), an audio circuit 160 (optional), a speaker 161 (optional), a microphone 162 (optional), a processor 170, an external interface 180, and a power supply 190. Those skilled in the art will appreciate that FIG. 1D is merely an example of a terminal or a multi-function device, and does not constitute a limitation on the terminal or the multi-function device, and may include more or fewer components than shown in the figure, or combine certain components, or different components.

The input unit 130 can be used to receive input digital or character information, and generate key signal input related to the user settings and function control of the portable multifunctional device. Specifically, the input unit 130 may include a touch screen 131 (optional) and/or other input devices 132. The touch screen 131 can collect the user's touch operations on or near it (such as the user's operation on or near the touch screen using any suitable object such as fingers, joints, stylus, etc.), and drive the corresponding connection device according to a pre-set program. The touch screen can detect the user's touch action on the touch screen, convert the touch action into a touch signal and send it to the processor 170, and can receive and execute the command sent by the processor 170; the touch signal at least includes the touch point coordinate information. The touch screen 131 can provide an input interface and an output interface between the terminal 100 and the user. In addition, the touch screen can be implemented using multiple types such as resistive, capacitive, infrared and surface acoustic wave. In addition to the touch screen 131, the input unit 130 can also include other input devices. Specifically, other input devices 132 may include, but are not limited to, one or more of a physical keyboard, function keys (such as a volume control key, a switch key, etc.), a trackball, a mouse, a joystick, and the like.

The input device 132 can receive input images, texts, and the like.

The display unit 140 may be used to display information input by the user or provided to the user, various menus of the terminal 100, interactive interfaces, file display, and/or playback of any multimedia file. In the embodiment of the present application, the display unit 140 may be used to display the interface of an image processing application, processing results, etc.

The memory 120 can be used to store instructions and data. The memory 120 can mainly include an instruction storage area and a data storage area. The data storage area can store various data, such as multimedia files, texts, etc.; the instruction storage area can store software units such as operating systems, applications, instructions required for at least one function, or their subsets and extensions. It can also include a non-volatile random access memory; provide the processor 170 with hardware, software and data resources including management of computing and processing equipment, and support control software and applications. It is also used for the storage of multimedia files, and the storage of running programs and applications.

The processor 170 is the control center of the terminal 100. It uses various interfaces and lines to connect various parts of the entire terminal 100. By running or executing instructions stored in the memory 120 and calling data stored in the memory 120, it executes various functions of the terminal 100 and processes data, thereby controlling the terminal device as a whole. Optionally, the processor 170 may include one or more processing units; preferably, the processor 170 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface and application program, and the modem processor mainly processes wireless communication. It is understandable that the above-mentioned modem processor may not be integrated into the processor 170. In some embodiments, the processor and the memory may be implemented on a single chip, and in some embodiments, they may also be implemented separately on separate chips. The processor 170 may also be used to generate corresponding operation control signals, send them to corresponding components of the computing and processing device, read and process data in the software, especially read and process data and programs in the memory 120, so that each functional module therein performs corresponding functions, thereby controlling the corresponding components to act according to the requirements of the instructions.

Among them, the memory 120 can be used to store software codes related to the data processing method, the processor 170 can execute the steps of the chip data processing method, and can also schedule other units (such as the above-mentioned input unit 130 and display unit 140) to realize corresponding functions.

The RF unit 110 (optional) can be used for receiving and sending information or receiving and sending signals during a call, for example, after receiving the downlink information of the base station, it is sent to the processor 170 for processing; in addition, the designed uplink data is sent to the base station. Generally, the RF circuit includes but is not limited to an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier (Low Noise Amplifier, LNA), a duplexer, etc. In addition, the RF unit 110 can also communicate with network devices and other devices through wireless communication. The wireless communication can use any communication standard or protocol, including but not limited to Global System of Mobile communication (GSM), General Packet Radio Service (GPRS), Code Division Multiple Access (CDMA), Wideband Code Division Multiple Access (WCDMA), Long Term Evolution (LTE), email, Short Messaging Service (SMS), etc.

In this embodiment of the present application, the RF unit 110 can send the image to the server 200 and receive the processing result sent by the server 200.

It should be understood that the radio frequency unit 110 is optional and can be replaced by other communication interfaces, such as a network port.

The terminal 100 also includes a power supply 190 (such as a battery) for supplying power to various components. Preferably, the power supply can be logically connected to the processor 170 through a power management system, so that the power management system can manage functions such as charging, discharging, and power consumption.

The terminal 100 also includes an external interface 180, which can be a standard Micro USB interface or a multi-pin connector. It can be used to connect the terminal 100 to communicate with other devices, and can also be used to connect a charger to charge the terminal 100.

Although not shown, the terminal 100 may also include a flashlight, a wireless fidelity (WiFi) module, a Bluetooth module, sensors with different functions, etc., which are not described in detail here. Some or all of the methods described below may be applied to the terminal 100 as shown in FIG. 1D .

Next, the product form of the server 200 in FIG. 1C is described;

Fig. 2 provides a schematic diagram of the structure of a server 200. As shown in Fig. 2, the server 200 includes a bus 201, a processor 202, a communication interface 203 and a memory 204. The processor 202, the memory 204 and the communication interface 203 communicate with each other via the bus 201.

The bus 201 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, FIG2 only uses a thick line, but does not mean that there is only one bus or one type of bus.

The processor 202 may be any one or more of a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).

The memory 204 may include a volatile memory (volatile memory), such as a random access memory (RAM). The memory 204 may also include a non-volatile memory (non-volatile memory), such as a read-only memory (ROM), a flash memory, a hard drive (HDD), or a solid state drive (SSD).

The memory 204 may be used to store software codes related to the data processing method, and the processor 202 may execute the steps of the data processing method of the chip, and may also schedule other units to implement corresponding functions.

It should be understood that the above-mentioned terminal 100 and server 200 can be centralized or distributed devices, and the processors in the above-mentioned terminal 100 and server 200 (such as processor 170 and processor 202) can be hardware circuits (such as application specific integrated circuit (ASIC), field-programmable gate array (FPGA), general-purpose processor, digital signal processor (DSP), microprocessor or microcontroller, etc.), or a combination of these hardware circuits. For example, the processor can be a hardware system with an instruction execution function, such as a CPU, DSP, etc., or a hardware system without an instruction execution function, such as an ASIC, FPGA, etc., or a combination of the above-mentioned hardware systems without an instruction execution function and hardware systems with an instruction execution function.

It should be understood that the steps related to the model reasoning process in the embodiments of the present application involve AI-related operations. When performing AI operations, the instruction execution architecture of the terminal device and the server is not limited to the processor combined with the memory architecture described above. The system architecture provided in the embodiments of the present application is described in detail below in conjunction with Figure 3.

FIG3 is a schematic diagram of a system architecture provided by an embodiment of the present application. As shown in FIG3 , a system architecture 500 includes an execution device 510 , a training device 520 , a database 530 , a client device 540 , a data storage system 550 , and a data acquisition system 560 .

The execution device 510 includes a calculation module 511, an I/O interface 512, a preprocessing module 513 and a preprocessing module 514. The calculation module 511 may include a target model/rule 501, and the preprocessing module 513 and the preprocessing module 514 are optional.

The execution device 510 may be a terminal device or a server that runs the above-mentioned image processing application.

The data acquisition device 560 is used to collect training samples. The training samples may be images and texts, etc. After collecting the training samples, the data acquisition device 560 stores the training samples in the database 530.

The training device 520 can train the neural network based on the training samples maintained in the database 530 to obtain the target model/rule 501.

It should be understood that the training device 520 can perform a pre-training process on the neural network to be trained based on the training samples maintained in the database 530, or fine-tune the model based on the pre-training.

It should be noted that, in actual applications, the training samples maintained in the database 530 may not all come from the data acquisition device 560, but may also be received from other devices. It should also be noted that the training device 520 may not train the target model/rule 501 entirely based on the training samples maintained in the database 530, but may also obtain training samples from the cloud or other places for model training. The above description should not be used as a limitation on the embodiments of the present application.

The target model/rule 501 trained by the training device 520 can be applied to different systems or devices, such as the execution device 510 shown in FIG3 . The execution device 510 can be a terminal, such as a mobile phone terminal, a tablet computer, a laptop computer, an augmented reality (AR)/virtual reality (VR) device, a vehicle-mounted terminal, etc., and can also be a server, etc.

Specifically, the training device 520 may transfer the trained model to the execution device 510 .

In Figure 3, the execution device 510 is configured with an input/output (I/O) interface 512 for data interaction with an external device. The user can input data (such as images and text in the embodiment of the present application) into the I/O interface 512 through the client device 540.

The preprocessing module 513 and the preprocessing module 514 are used to preprocess the input data received by the I/O interface 512. It should be understood that there may be no preprocessing module 513 and the preprocessing module 514 or only one preprocessing module. When there is no preprocessing module 513 and the preprocessing module 514, the computing module 511 may be directly used to process the input data.

When the execution device 510 preprocesses the input data, or when the computing module 511 of the execution device 510 performs calculations and other related processing, the execution device 510 can call the data, code, etc. in the data storage system 550 for corresponding processing, and can also store the data, instructions, etc. obtained from the corresponding processing into the data storage system 550.

Finally, the I/O interface 512 provides the processing results to the client device 540 and thus to the user.

In the case shown in FIG. 3 , the user can manually give input data, and the “manually given input data” can be operated through the interface provided by the I/O interface 512. In another case, the client device 540 can automatically send input data to the I/O interface 512. If the client device 540 is required to automatically send input data and needs to obtain the user's authorization, the user can set the corresponding authority in the client device 540. The user can view the results output by the execution device 510 on the client device 540, and the specific presentation form can be a specific method such as display, sound, action, etc. The client device 540 can also be used as a data acquisition terminal to collect the input data of the input I/O interface 512 and the output results of the output I/O interface 512 as shown in the figure as new sample data, and store them in the database 530. Of course, it is also possible not to collect through the client device 540, but the I/O interface 512 directly stores the input data of the input I/O interface 512 and the output results of the output I/O interface 512 as new sample data in the database 530.

It is worth noting that FIG. 3 is only a schematic diagram of a system architecture provided by an embodiment of the present application, and the positional relationship between the devices, components, modules, etc. shown in the figure does not constitute any limitation. For example, in FIG. 3, the data storage system 550 is an external memory relative to the execution device 510. In other cases, the data storage system 550 can also be placed in the execution device 510. It should be understood that the above-mentioned execution device 510 can be deployed in the client device 540.

From the inference side of the model:

In the embodiment of the present application, the computing module 511 of the above-mentioned execution device 520 can obtain the code stored in the data storage system 550 to implement the steps related to the model reasoning process in the embodiment of the present application.

In the embodiment of the present application, the computing module 511 of the execution device 520 may include a hardware circuit (such as an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a general-purpose processor, a digital signal processor (DSP), a microprocessor or a microcontroller, etc.), or a combination of these hardware circuits. For example, the training device 520 can be a hardware system with an execution instruction function, such as a CPU, a DSP, etc., or a hardware system without an execution instruction function, such as an ASIC, an FPGA, etc., or a combination of the above-mentioned hardware systems without an execution instruction function and hardware systems with an execution instruction function.

Specifically, the computing module 511 of the execution device 520 can be a hardware system with an execution instruction function, and the steps related to the model reasoning process provided in the embodiment of the present application can be software codes stored in the memory. The computing module 511 of the execution device 520 can obtain the software code from the memory and execute the obtained software code to implement the steps related to the model reasoning process provided in the embodiment of the present application.

It should be understood that the computing module 511 of the execution device 520 can be a combination of a hardware system that does not have the function of executing instructions and a hardware system that has the function of executing instructions. Some of the steps related to the model reasoning process provided in the embodiments of the present application can also be implemented by the hardware system that does not have the function of executing instructions in the computing module 511 of the execution device 520, which is not limited here.

From the training side of the model:

In an embodiment of the present application, the above-mentioned training device 520 can obtain the code stored in the memory (not shown in Figure 3, which can be integrated into the training device 520 or deployed separately from the training device 520) to implement the steps related to model training in an embodiment of the present application.

In the embodiment of the present application, the training device 520 may include a hardware circuit (such as an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a general processor, a digital signal processor, or a processor. For example, the training device 520 may be a hardware system with the function of executing instructions, such as a CPU, DSP, etc., or a hardware system without the function of executing instructions, such as an ASIC, FPGA, etc., or a combination of the above-mentioned hardware systems without the function of executing instructions and hardware systems with the function of executing instructions.

It should be understood that the training device 520 can be a combination of a hardware system that does not have the function of executing instructions and a hardware system that has the function of executing instructions. Some of the steps related to model training provided in the embodiments of the present application can also be implemented by the hardware system that does not have the function of executing instructions in the training device 520, which is not limited here.

2. Image processing cloud services provided by the server:

In one possible implementation, the server can provide image processing services to the end side through an application programming interface (API).

Among them, the terminal device can send relevant parameters (such as images and texts) to the server through the API provided by the cloud. The server can obtain processing results based on the received parameters, etc., and return the processing results to the terminal.

The description of the terminal and the server can be the same as that of the above embodiments, and will not be repeated here.

FIG. 4 shows a process of using an image processing cloud service provided by a cloud platform.

1. Activate and purchase content review service.

2. Users can download the software development kit (SDK) corresponding to the content review service. Usually, the cloud platform provides multiple development versions of the SDK for users to choose according to the requirements of the development environment, such as JAVA version SDK, Python version SDK, PHP version SDK, Android version SDK, etc.

3. After the user downloads the corresponding version of the SDK to the local computer according to the needs, the SDK project is imported into the local development environment, and configuration and debugging are performed in the local development environment. The local development environment can also be used to develop other functions, thus forming an application that integrates image processing capabilities.

4. When image processing applications are used, they can trigger an API call for image processing when image processing is required. When the application triggers the image processing function, it initiates an API request to the running instance of the image processing service in the cloud environment. The API request carries the image, and the running instance in the cloud environment processes the image to obtain the processing result.

5. The cloud environment returns the processing results to the application, thereby completing a method call provided in an embodiment of the present application.

Since the embodiments of the present application involve the application of a large number of neural networks, in order to facilitate understanding, the relevant terms and related concepts such as neural networks involved in the embodiments of the present application are first introduced below.

(1) Neural Network

A neural network may be composed of neural units, and a neural unit may refer to an operation unit that takes _xs (i.e., input data) and intercept 1 as input, and the output of the operation unit may be:

Where s=1, 2, ...n, n is a natural number greater than 1, Ws is the weight of xs, and b is the bias of the neural unit. f is the activation function of the neural unit, which is used to introduce nonlinear characteristics into the neural network to convert the input signal in the neural unit into the output signal. The output signal of the activation function can be used as the input of the next convolutional layer, and the activation function can be a sigmoid function. A neural network is a network formed by connecting multiple single neural units mentioned above, that is, the output of one neural unit can be the input of another neural unit. The input of each neural unit can be connected to the local receptive field of the previous layer to extract the characteristics of the local receptive field. The local receptive field can be an area composed of several neural units.

(2) Convolutional neural network (CNN) is a deep neural network with a convolutional structure. Convolutional neural network contains a feature extractor consisting of a convolution layer and a subsampling layer, which can be regarded as a filter. The convolution layer refers to the neuron layer in the convolutional neural network that performs convolution processing on the input signal. In the convolution layer of the convolutional neural network, a neuron can only be connected to some neurons in the adjacent layers. A convolutional layer usually contains several feature planes, and each feature plane can be composed of some neural units arranged in a rectangular shape. The neural units in the same feature plane share weights, and the shared weights here are the convolution kernels. Shared weights It can be understood that the way to extract features is independent of position. The convolution kernel can be formalized as a matrix of random size, and the convolution kernel can obtain reasonable weights through learning during the training process of the convolutional neural network. In addition, the direct benefit of shared weights is to reduce the connections between the layers of the convolutional neural network, while reducing the risk of overfitting.

CNN is a very common neural network. The following is a detailed introduction to the structure of CNN in conjunction with Figure 5A. As mentioned in the previous basic concept introduction, convolutional neural network is a deep neural network with a convolution structure and a deep learning architecture. Deep learning architecture refers to multiple levels of learning at different abstract levels through machine learning algorithms. As a deep learning architecture, CNN is a feed-forward artificial neural network in which each neuron can respond to the image input into it.

As shown in Figure 5A, the convolutional neural network (CNN) 200 may include an input layer 210, a convolutional layer/pooling layer 220 (wherein the pooling layer is optional), and a fully connected layer (fully connected layer) 230.

Convolutional layer/pooling layer 220:

Convolutional Layer:

As shown in FIG5A , the convolution layer/pooling layer 220 may include layers 221-226, for example: in one implementation, layer 221 is a convolution layer, layer 222 is a pooling layer, layer 223 is a convolution layer, layer 224 is a pooling layer, layer 225 is a convolution layer, and layer 226 is a pooling layer; in another implementation, layers 221 and 222 are convolution layers, layer 223 is a pooling layer, layers 224 and 225 are convolution layers, and layer 226 is a pooling layer. That is, the output of a convolution layer can be used as the input of a subsequent pooling layer, or as the input of another convolution layer to continue the convolution operation.

The following will take the convolution layer 221 as an example to introduce the internal working principle of a convolution layer.

The convolution layer 221 may include a plurality of convolution operators, which are also called kernels. The convolution operator is equivalent to a filter that extracts specific information from the input image matrix in image processing. The convolution operator may be essentially a weight matrix, which is usually predefined. In the process of performing convolution operations on the image, the weight matrix is usually processed one pixel after another (or two pixels after two pixels... depending on the value of the stride) in the horizontal direction on the input image, thereby completing the work of extracting specific features from the image. The size of the weight matrix should be related to the size of the image. It should be noted that the depth dimension of the weight matrix is the same as the depth dimension of the input image. In the process of performing convolution operations, the weight matrix extends to the entire depth of the input image. Therefore, convolution with a single weight matrix will produce a convolution output with a single depth dimension, but in most cases, a single weight matrix is not used, but multiple weight matrices of the same size (row × column), that is, multiple isotype matrices, are applied. The output of each weight matrix is stacked to form the depth dimension of the convolved image, and the dimension here can be understood as being determined by the "multiple" mentioned above. Different weight matrices can be used to extract different features in the image, for example, one weight matrix is used to extract image edge information, another weight matrix is used to extract specific colors of the image, and another weight matrix is used to blur unwanted noise in the image, etc. The multiple weight matrices have the same size (rows × columns), and the feature maps extracted by the multiple weight matrices of the same size are also the same size. The extracted feature maps of the same size are then merged to form the output of the convolution operation.

The weight values in these weight matrices need to be obtained through a lot of training in practical applications. The weight matrices formed by the weight values obtained through training can be used to extract information from the input image, so that the convolutional neural network 200 can make correct predictions.

When the convolutional neural network 200 has multiple convolutional layers, the initial convolutional layer (for example, 221) often extracts more general features, which can also be called low-level features. As the depth of the convolutional neural network 200 increases, the features extracted by the later convolutional layers (for example, 226) become more and more complex, such as high-level semantic features. Features with higher semantics are more suitable for the problem to be solved.

Pooling layer:

Since it is often necessary to reduce the number of training parameters, it is often necessary to periodically introduce a pooling layer after the convolution layer. In the layers 221-226 shown in 220 in FIG. 5A, a convolution layer may be followed by a pooling layer, or multiple convolution layers may be followed by one or more pooling layers. In the image processing process, the only purpose of the pooling layer is to reduce the spatial size of the image. The pooling layer may include an average pooling operator and/or a maximum pooling operator to sample the input image to obtain an image of smaller size. The average pooling operator may calculate the pixel values in the image within a specific range to generate an average value as the result of average pooling. The maximum pooling operator may take the pixel with the largest value in the range within a specific range as the result of maximum pooling. In addition, just as the size of the weight matrix used in the convolution layer should be related to the image size, the operator in the pooling layer should also be related to the image size. The size of the image output after processing by the pooling layer may be smaller than the size of the image input to the pooling layer, and each pixel in the image output by the pooling layer represents the average value or maximum value of the corresponding sub-region of the image input to the pooling layer.

Fully connected layer 230:

After being processed by the convolution layer/pooling layer 220, the convolution neural network 200 is not sufficient to output the required output information. Because as mentioned above, the convolution layer/pooling layer 220 will only extract features and reduce the parameters brought by the input image. However, in order to generate the final output information (the required class information or other related information), the convolution neural network 200 needs to use the fully connected layer 230 to generate one or a group of outputs of the required number of classes. Therefore, the fully connected layer 230 may include multiple hidden layers (231, 232 to 23n as shown in Figure 5A), and the parameters contained in the multiple hidden layers can be pre-trained according to the relevant training data of the specific task type. For example, the task type may include image recognition, image classification, image super-resolution reconstruction, etc.

After the multiple hidden layers in the fully connected layer 230, that is, the last layer of the entire convolutional neural network 200 is the output layer 240, which has a loss function similar to the classification cross entropy, and is specifically used to calculate the prediction error. Once the forward propagation of the entire convolutional neural network 200 (such as the propagation from 210 to 240 in Figure 5A is forward propagation) is completed, the back propagation (such as the propagation from 240 to 210 in Figure 5A is back propagation) will begin to update the weight values and biases of the aforementioned layers to reduce the loss of the convolutional neural network 200 and the error between the result output by the convolutional neural network 200 through the output layer and the ideal result.

It should be noted that the convolutional neural network 200 shown in FIG. 5A is only an example of a convolutional neural network. In specific applications, the convolutional neural network may also exist in the form of other network models, for example, including only a part of the network structure shown in FIG. 5A. For example, the convolutional neural network used in the embodiment of the present application may only include an input layer 210, a convolutional layer/pooling layer 220 and an output layer 240.

It should be noted that the convolutional neural network 100 shown in FIG. 5A is only an example of a convolutional neural network. In specific applications, the convolutional neural network can also exist in the form of other network models. For example, as shown in FIG. 5B , multiple convolutional layers/pooling layers are used in parallel, and the features extracted respectively are input to the fully connected layer 230 for processing.

(3) Deep Neural Networks

Deep Neural Network (DNN), also known as multi-layer neural network, can be understood as a neural network with many hidden layers. There is no special metric for "many" here. From the position of different layers of DNN, the neural network inside DNN can be divided into three categories: input layer, hidden layer, and output layer. Generally speaking, the first layer is the input layer, the last layer is the output layer, and the layers in between are all hidden layers. The layers are fully connected, that is, any neuron in the i-th layer must be connected to any neuron in the i+1-th layer. Although DNN looks complicated, the work of each layer is actually not complicated. Simply put, it is the following linear relationship expression: in, is the input vector, is the output vector, is the offset vector, W is the weight matrix (also called coefficient), and α() is the activation function. Each layer is just an input vector After such a simple operation, the output vector Since DNN has many layers, the coefficient W and the offset vector The definition of these parameters in DNN is as follows: Take the coefficient W as an example: Assume that in a three-layer DNN, the linear coefficient from the 4th neuron in the second layer to the 2nd neuron in the third layer is defined as The superscript 3 represents the layer number of coefficient W, while the subscripts correspond to the third layer index 2 of the output and the second layer index 4 of the input.

In summary, the coefficients from the kth neuron in the L-1th layer to the jth neuron in the Lth layer are defined as

It should be noted that the input layer does not have a W parameter. In a deep neural network, more hidden layers allow the network to better describe complex situations in the real world. Theoretically, the more parameters a model has, the higher its complexity and the greater its "capacity", which means it can complete more complex learning tasks. Training a deep neural network is the process of learning the weight matrix, and its ultimate goal is to obtain the weight matrix of all layers of the trained deep neural network (a weight matrix formed by many layers of vectors W).

(4) Loss Function

In the process of training a deep neural network, because we hope that the output of the deep neural network is as close as possible to the value we really want to predict, we can compare the predicted value of the current network with the target value we really want, and then update the weight vector of each layer of the neural network according to the difference between the two (of course, there is usually an initialization process before the first update, that is, pre-configuring parameters for each layer in the deep neural network). For example, if the predicted value of the network is high, adjust the weight vector to make it predict a lower value, and keep adjusting until the deep neural network can predict the target value we really want or a value very close to the target value we really want. Therefore, it is necessary to pre-define "how to compare the difference between the predicted value and the target value", which is the loss function or objective function, which are important equations used to measure the difference between the predicted value and the target value. Among them, taking the loss function as an example, the higher the output value (loss) of the loss function, the greater the difference, so the training of the deep neural network becomes a process of minimizing this loss as much as possible.

(5) Back propagation algorithm

Convolutional neural networks can use the back propagation (BP) algorithm to correct the size of the parameters in the initial super-resolution model during the training process, so that the reconstruction error loss of the super-resolution model becomes smaller and smaller. Error loss will be generated, and the parameters in the initial super-resolution model are updated by back-propagating the error loss information, so that the error loss converges. The back-propagation algorithm is a back-propagation movement dominated by error loss, aiming to obtain the optimal parameters of the super-resolution model, such as the weight matrix.

(6) Attention mechanism

The attention mechanism imitates the internal process of biological observation behavior, that is, a mechanism that aligns internal experience and external sensations to increase the observation precision of some areas, and can use limited attention resources to quickly filter out high-value information from a large amount of information. The attention mechanism can quickly extract important features of sparse data, and is therefore widely used in natural language processing tasks, especially machine translation. The self-attention mechanism is an improvement on the attention mechanism, which reduces dependence on external information and is better at capturing the internal correlation of data or features. The essential idea of the attention mechanism can be rewritten as the following formula:

Among them, Lx＝||Source|| represents the length of Source. The formula means that the elements in Source are imagined to be composed of a series of data pairs. At this time, given a certain element Query in the target Target, by calculating the similarity or correlation between Query and each Key, the weight coefficient of the Value corresponding to each Key is obtained, and then the Value is weighted and summed to obtain the final Attention value. Therefore, the Attention mechanism is essentially a weighted sum of the Value values of the elements in Source, and Query and Key are used to calculate the weight coefficient of the corresponding Value. Conceptually, Attention can be understood as selectively filtering out a small amount of important information from a large amount of information and focusing on these important information, ignoring most of the unimportant information. The focusing process is reflected in the calculation of the weight coefficient. The larger the weight, the more focus is placed on its corresponding Value value, that is, the weight represents the importance of the information, and the Value is the corresponding information. The self-attention mechanism can be understood as internal Attention (intra attention). The Attention mechanism occurs between the Query element of the Target and all the elements in the Source. The self-attention mechanism refers to the Attention mechanism that occurs between the internal elements of the Source or between the internal elements of the Target. It can also be understood as the attention calculation mechanism in the special case of Target=Source. The specific calculation process is the same, but the calculation object has changed.

(7) Grounding data: visual positioning data (including pictures and corresponding descriptions). There are multiple boxes in a picture, and each box corresponds to a phrase in the description, describing the object or state in the box.

(8) Detection data: Conventional detection data, there are multiple boxes in an image, and one box corresponds to a noun category.

In the current field of machine vision, open domain detection models, such as DetCLIP, have demonstrated powerful capabilities and can achieve accurate target detection for any given category. However, this model is not impeccable, and there are some obvious problems in its practical application. First, we often cannot provide some specific object categories. This is because there are many kinds of objects in the real world, but our knowledge base and vocabulary are limited. Secondly, the category tables we list are often incomplete. This problem is more due to the limitations of our knowledge acquisition and organization methods. Finally, we cannot give specific categories based on some attributes, functions and other knowledge of the category. This is because the current machine vision system lacks the ability to deeply understand and comprehensively apply knowledge.

However, there are already some large language models, such as ChatGPT, and multimodal large models, such as miniGPT4, which have excellent knowledge understanding and reasoning capabilities. These models can understand and process human natural language, and can also make meaningful inferences based on given information. But they also have some shortcomings, such as the lack of predictive capabilities for fine-grained tasks such as object detection.

In order to solve the above problems, refer to Figure 6, which is a flow chart of a data processing method provided in an embodiment of the present application. As shown in Figure 6, a data processing method provided in an embodiment of the present application may include steps 601 to 603, and these steps are described in detail below.

601. Obtain an image and an object detection request, where the object detection request includes a natural language description of an object to be detected in the image.

Among them, steps 601 to 603 can be the feedforward process of the model training process, or can be the reasoning process of the model.

In an embodiment of the present application, it is desired to detect and output object categories and their positions (that is, the position of the detection box in the image) that meet the user's language conditions when a user inputs a language format and a picture of the current scene.

During the training process of the model, training samples including images and object detection requests (also referred to as human language instructions) can be obtained.

For example, a labeling process can be built based on the existing object detection dataset and language model. Through this data labeling process, We have built an open domain detection dataset based on human language instructions. In this dataset, human language instructions are refined and divided into the following four tasks according to characteristics and requirements:

Task 1 is to detect all categories. In this task, we hope that the model can fully and thoroughly detect and determine all object categories in the image after receiving the user's input.

Task 2 is to detect partial categories. Compared with the comprehensive detection of Task 1, this task requires the model to be more targeted to identify and detect specific object categories specified by the user.

Task 3 is to detect the categories contained in a parent category. In this task, the model needs to understand and identify the hierarchical relationship between objects. For example, if the user asks to detect all fruit categories, the model needs to detect all fruits in the picture, such as apples, oranges, etc.

Task 4 is to detect categories with a certain function or attribute. This task requires the model to have a higher level of understanding and recognition capabilities. For example, when the user instructs to detect all objects that can be used for writing, the model needs to recognize different categories of objects such as pencils, pens, and markers.

In this way, the model can better understand and execute the user's language instructions, thereby more accurately identifying and locating objects in images.

For example, referring to FIG. 7A , FIG. 7A is a schematic diagram of a natural language description of an object detection request for each task in a training sample.

During the inference process of the model, an object detection request may indicate target detection on an image, and the semantics of the object detection request includes a description of the object to be detected.

602. Process the image by an image encoder to obtain a feature representation of each of a plurality of image regions, each image region corresponding to a candidate detection frame;

In the existing implementation, when performing target detection, the image encoder outputs the feature representation of the entire image, and the related training method makes the language model not have/not good at fine-grained image processing capabilities. In the embodiment of the present application, the image encoder outputs the feature representation of each of the multiple image areas in the image, and each image area can correspond to a candidate detection box, and each detection box can contain an object. The area where the object to be detected indicated in the object detection request is located can be selected from multiple image areas. Specifically, the feature representation of each image area and the object detection request can be input into the language model, and the language model determines whether the object included in each image area is the object to be identified indicated by the object detection request. The language model can also output the category of the object included in the image area when it is determined that the object included in the image area is the object to be identified indicated by the object detection request.

Using the image encoder to obtain the features of each image area and combining it with the language model for target detection can improve the processing capabilities for fine-grained features and improve the accuracy of target detection under the guidance of human language.

Next, we will introduce how to train an image encoder with the above-mentioned fine-grained processing capabilities.

In a possible implementation, a training sample of an image encoder can be obtained, wherein the training sample includes an image sample and a corresponding text sample, wherein the text sample is a text description corresponding to the image sample; the image sample is processed by the image encoder to obtain a first processing result; the text sample is processed by the text encoder to obtain a second processing result; and the image encoder and the text encoder are updated through comparative learning based on the first processing result and the second processing result.

In a possible implementation, the first processing result includes multiple detection boxes and a category of each detection box; the image encoder and the text encoder can be updated according to the first processing result and the corresponding true value.

In terms of model architecture, for image encoders, the model can use the same or similar architecture as DetCLIP. Specifically, the same ATSS architecture as the swin-T backbone can be used. For text encoders, the FlanT5 model family can be used, for example, it can be performed on OPT, FlanT5-base, and FlanT5-Large at the same time.

Exemplarily, the training process of the embodiment of the present application may include two stages. Referring to Figure 7B, the above-mentioned training process for the image encoder and the text encoder may belong to the first stage. The first stage is similar to the pre-training of the open vocabulary object detector. For example, training data from detection, positioning and captioning tasks may be used. In the first stage, the DetCLIP method can be used to align features with text. Specifically, open vocabulary detection pre-training can be performed in the DetCLIP manner. The detector is pre-trained by optimizing the fine contrast loss between the text embedding and the object-level visual embedding, as well as the center loss and the bounding box regression loss. Visual-text feature alignment is performed using detection, positioning, and image-text pair datasets. After the first stage, a vocabulary object detector is obtained, which can extract visual object embeddings that are well aligned with text embeddings from the pre-trained CLIP text encoder.

For example, in the first stage, the training method of DetCLIP can be continued. Given an input image x, firstly, M RoI region features V ⁱ , i∈[1,M] are obtained through an image encoder Φ _i (for example, an ATSS single-stage detector can be used), and the single-stage detector is calculated. The center loss L _CEN (sigmoid cross entropy loss), regression loss L _REG (giou loss) and alignment loss function _LALI (alignment loss) corresponding to the test model are used. Note that for detection or grounding data, large resolution and small batch size can be used for training, and for image-text data, small resolution and large batch size can be used for training.

603. Process the object detection request and the multiple feature representations through a language model, and determine a region where the object to be detected is located from the multiple image regions.

In one possible implementation, a target detection result of the corresponding image area can be obtained based on each feature representation and object detection request. The detection result can indicate whether the object included in the image area is the object indicated in the object detection request, and the category corresponding to the object when the object included in the image area is the object indicated in the object detection request.

In a possible implementation, referring to FIG7B , the language model may output “other” when determining that the object included in the image region is not the object indicated in the object detection request. That is, only the detection boxes that meet the instructions will be classified, and the others are marked as other categories.

In a possible implementation, the object detection request and each of the feature representations can be processed in parallel through a language model to obtain a detection result for each of the image regions; and based on the detection result, the region where the object to be detected is located is determined from among the multiple image regions.

That is, the language model can determine the detection results for each image region in parallel.

After extracting object-level visual features from an image using DetCLIP, there are two ways to train a model to achieve the goals of an embodiment of the present application. The first method is to connect the object features and train a language model to sequentially predict the output of each object. However, this does not conform to the original output habits of the language long-term memory model (LLM), which increases the difficulty of training. Another method is to allow the language model to process each object feature independently. Specifically, the box-based object features obtained by each detection model interact with the corresponding instructions through LLM, and then LLM only predicts the specific category output of the object based on the instruction.

In a possible implementation, the language model includes multiple network layers, at least one of the multiple network layers includes a cross-attention layer; the cross-attention layer is used to perform attention interaction between image features and text features.

In the model training process, the training of the language model can be the second stage. In the second stage, other parameters except the language model can be fixed, and the language model (such as the feature interaction module in the language model) can be trained to realize the image and text understanding detection under human instructions. Specifically:

In the second stage, the object detector is given the ability to follow human instructions by introducing a language model into the model. For example, the model can be trained on the IOD-Bench training set and predict object category names only for those objects that meet the instructions.

In the second stage, instruction adjustment is performed using the training data in IOD-Bench. Specifically, the image encoder and pre-trained language model obtained in the first stage are frozen. Then, a randomly initialized cross-attention layer (that is, the cross-attention layer in the embodiment of the present application) is inserted into the decoding layer of the language model, and training is started from scratch. The image is processed by the image encoder to extract object-level visual features. At the same time, the accompanying text instructions are processed by the language model. Cross-attention operations are performed between object-level visual features and text features. Finally, the language modeling loss output by the language model is optimized.

For example, given an input image x, firstly, M object-level visual features V ⁱ , i∈[1,M] are obtained through an image encoder Φ _i (for example, an ATSS single-stage detector can be used). In order to achieve cross-modal fusion, a randomly initialized cross-attention layer is inserted into the decoder layer of the language model and training is started from scratch. For each RoI feature and input text T from the visual part, the operation on each block of the LLM layer is:
h _L+1 = FF [tanh (α) × X Attn (Attn (h _L ), V) + Attn (h _L )];

Where FF, XAttn and Attn refer to feed forward network, cross attention layer and self-attention layer respectively. hL is the input of text features corresponding to the Lth layer block, and a is a learnable parameter initialized to 0.

Finally, the language modeling loss that optimizes the output of the language model can be formulated as follows:

Where Φ represents LLM, ^Vi is the visual feature of the i-th object, is the text token associated with the ith object at the tth time step, Refers to the text token related to the i-th object before the t-th time.

In the embodiment of the present application, a fine-grained feature interaction module based on object level is proposed. Different from the traditional full-image feature input, the model introduces object-level visual features, so that the feature interaction module can learn the fusion of object and instruction features. The processing of fine-grained features is further optimized.

Although the prior art DetCLIP has shown excellent performance in many indicators, it also has some obvious limitations. First, it needs to give a specific object category, which may not be convenient in some scenarios. Secondly, it cannot directly detect according to human language instructions, which limits its practicality in applications such as human-computer interaction. The embodiment of the present application introduces a large language model to increase the model's understanding and reasoning ability of pictures. The specific approach includes: inserting a randomly initialized cross-attention layer in the decoding layer of the language model and training from scratch. The image is processed by the encoder to extract object-level visual features. At the same time, the accompanying text instructions are processed by the language model. Cross-attention operations are performed between object-level visual features and text features. Finally, the language modeling loss output by the language model is optimized.

Next, the beneficial effects of the embodiments of the present application are introduced in combination with specific experiments:

By comparing with the baseline method on the IOD-Bench dataset, the effectiveness of the model and training strategy of the embodiment of the present application is demonstrated. It is compared with variants of BLIP2 and MiniGPT4. For BLIP2, experiments were conducted using FlanT5-XL and FlanT5-XXL as language models. For MiniGPT4, Vicunna-7b was used as the language model. As shown in Table 1, Ins-DetCLIP of the embodiment of the present application significantly surpasses other opponents. Specifically, Ins-DetCLIP using the FlanT5-base model has surpassed the MiniGPT4 baseline by an average of 9.78% on all tasks. Due to the generalization ability of LLM, the model of the embodiment of the present application also shows good performance on instructions that did not appear during training. For example, for invisible domain instructions, Ins-DetCLIP using FlanT5-base can still achieve an average mAP of 13.7, which is only slightly lower than the result on in-domain instructions by 1.6%.

Table 1

With the excellent generation ability of LLM, Ins-DetCLIP of the embodiment of the present application can not only predict category names, but also generate detailed descriptions for objects of interest. By benchmarking it on the Dense Captioning task, the superior description generation ability of the model of the embodiment of the present application is demonstrated. To ensure a fair comparison, the regression head of Ins-DetCLIP was fine-tuned using the box annotations of the Dense Captioning dataset. As shown in Table 2, the model of the embodiment of the present application consistently outperforms other methods and achieves state-of-the-art results.

Table 2

In terms of inference efficiency, the inference efficiency between Ins-DetCLIP and the two-stage baseline is compared in terms of frames per second (FPS). As shown in Table 3, compared with the baseline method, the model of the embodiment of the present application can achieve better performance and faster inference speed.

Table 3

Referring to FIG. 8 , FIG. 8 is a schematic diagram of the structure of a data processing device provided in an embodiment of the present application. As shown in FIG. 8 , a data processing device 800 provided in an embodiment of the present application includes:

An acquisition module 801 is used to acquire an image and an object detection request, wherein the object detection request includes a natural language description of an object to be detected in the image;

The description of the acquisition module 801 may refer to the description of step 601 in the above embodiment, which will not be repeated here.

A processing module 802 is used to process the image through an image encoder to obtain a feature representation of each image region in a plurality of image regions, each image region corresponding to a candidate detection frame;

The object detection request and the plurality of feature representations are processed by a language model, and a region where the object to be detected is located is determined from among the plurality of image regions.

The description of the processing module 802 may refer to the description of steps 602 and 603 in the above embodiment, which will not be repeated here.

In a possible implementation, the processing module 802 is specifically configured to:

Processing the object detection request and each of the feature representations in parallel through a language model to obtain a detection result for each of the image regions;

According to the detection result, the area where the object to be detected is located is determined from the multiple image areas.

In a possible implementation, before the image is processed by the image encoder, the acquisition module 801 is further configured to:

Acquire a training sample, wherein the training sample includes an image sample and a corresponding text sample, wherein the text sample is a text description corresponding to the image sample;

The processing module 802 is further configured to: process the image sample through an image encoder to obtain a first processing result;

Processing the text sample through a text encoder to obtain a second processing result;

According to the first processing result and the second processing result, the image encoder and the text encoder are updated through comparative learning.

In a possible implementation, the first processing result includes a plurality of detection frames and a category of each detection frame; and the processing module 802 is further configured to:

The image encoder and the text encoder are updated according to the first processing result and the corresponding true value.

In a possible implementation, the language model includes multiple network layers, at least one of the multiple network layers includes a cross-attention layer; the cross-attention layer is used to perform attention interaction between image features and text features.

Next, an execution device provided in an embodiment of the present application is introduced. Please refer to Figure 9. Figure 9 is a structural schematic diagram of an execution device provided in an embodiment of the present application. The execution device 900 can be specifically manifested as a virtual reality VR device, a mobile phone, a tablet, a laptop computer, a smart wearable device, a monitoring data processing device or a server, etc., which is not limited here. Specifically, the execution device 900 includes: a receiver 901, a transmitter 902, a processor 903 and a memory 904 (wherein the number of processors 903 in the execution device 900 can be one or more, and one processor is taken as an example in Figure 9), wherein the processor 903 may include an application processor 9031 and a communication processor 9032. In some embodiments of the present application, the receiver 901, the transmitter 902, the processor 903 and the memory 904 may be connected via a bus or other means.

The memory 904 may include a read-only memory and a random access memory, and provides instructions and data to the processor 903. A portion of the memory 904 may also include a non-volatile random access memory (NVRAM). The memory 904 stores processor and operation instructions, executable modules or data structures, or subsets thereof, or extended sets thereof, wherein the operation instructions may include various operation instructions for implementing various operations.

The processor 903 controls the operation of the execution device. In a specific application, the various components of the execution device are coupled together through a bus system, wherein the bus system may include a power bus, a control bus, and a status signal bus in addition to a data bus. However, for the sake of clarity, For the sake of explanation, various buses are referred to as bus systems in the figures.

The method disclosed in the above embodiment of the present application can be applied to the processor 903, or implemented by the processor 903. The processor 903 can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit in the processor 903 or the instruction in the form of software. The above processor 903 can be a general processor, a digital signal processor (digital signal processing, DSP), a microprocessor or a microcontroller, and can further include an application specific integrated circuit (application specific integrated circuit, ASIC), a field programmable gate array (field-programmable gate array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The processor 903 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiment of the present application. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to be executed, or a combination of hardware and software modules in the decoding processor can be executed. The software module may be located in a storage medium mature in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 904, and the processor 903 reads the information in the memory 904 and completes the steps of the above method in combination with its hardware.

The receiver 901 can be used to receive input digital or character information and generate signal input related to the relevant settings and function control of the execution device. The transmitter 902 can be used to output digital or character information through the first interface; the transmitter 902 can also be used to send instructions to the disk group through the first interface to modify the data in the disk group; the transmitter 902 can also include a display device such as a display screen.

The embodiment of the present application also provides a training device, please refer to Figure 10, Figure 10 is a structural diagram of the training device provided by the embodiment of the present application, specifically, the training device 1000 is implemented by one or more servers, and the training device 1000 may have relatively large differences due to different configurations or performances, and may include one or more central processing units (CPU) 1010 (for example, one or more processors) and memory 1032, one or more storage media 1030 (for example, one or more mass storage devices) storing application programs 1042 or data 1044. Among them, the memory 1032 and the storage medium 1030 can be short-term storage or permanent storage. The program stored in the storage medium 1030 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the training device. Furthermore, the central processor 1010 can be configured to communicate with the storage medium 1030 to execute a series of instruction operations in the storage medium 1030 on the training device 1000.

The training device 1000 may also include one or more power supplies 1026, one or more wired or wireless network interfaces 1050, one or more input and output interfaces 1058; or, one or more operating systems 1041, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

In the embodiment of the present application, the central processing unit 1010 is used to execute actions related to model training in the above embodiment.

Also provided in an embodiment of the present application is a computer program product which, when executed on a computer, enables the computer to execute the steps executed by the aforementioned execution device, or enables the computer to execute the steps executed by the aforementioned training device.

A computer-readable storage medium is also provided in an embodiment of the present application, which stores a program for signal processing. When the computer-readable storage medium is run on a computer, it enables the computer to execute the steps executed by the aforementioned execution device, or enables the computer to execute the steps executed by the aforementioned training device.

The execution device, training device or terminal device provided in the embodiments of the present application may specifically be a chip, and the chip includes: a processing unit and a communication unit, wherein the processing unit may be, for example, a processor, and the communication unit may be, for example, an input/output interface, a pin or a circuit, etc. The processing unit may execute the computer execution instructions stored in the storage unit so that the chip in the execution device executes the data processing method described in the above embodiment, or so that the chip in the training device executes the data processing method described in the above embodiment. Optionally, the storage unit is a storage unit in the chip, such as a register, a cache, etc. The storage unit may also be a storage unit located outside the chip in the wireless access device end, such as a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM), etc.

Specifically, please refer to FIG. 11, which is a schematic diagram of the structure of a chip provided in an embodiment of the present application. The chip can be expressed as a neural network processor NPU 1100. NPU 1100 is mounted on the host CPU (Host CPU) as a coprocessor, and tasks are assigned by the Host CPU. The core part of the NPU is the operation circuit 1103, which is controlled by the controller 1104 to extract matrix data from the memory and perform multiplication operations.

In some implementations, the operation circuit 1103 includes multiple processing units (Process Engine, PE) inside. In some implementations, the operation circuit 1103 is a two-dimensional systolic array. The operation circuit 1103 can also be a one-dimensional systolic array or other electronic circuits capable of performing mathematical operations such as multiplication and addition. In some implementations, the operation circuit 1103 is a general-purpose matrix processor.

For example, assume there is an input matrix A, a weight matrix B, and an output matrix C. The operation circuit takes the corresponding data of matrix B from the weight memory 1102 and caches it on each PE in the operation circuit. The operation circuit takes the matrix A data from the input memory 1101 and performs matrix operation with matrix B, and the partial result or final result of the matrix is stored in the accumulator 1108.

The unified memory 1106 is used to store input data and output data. The weight data is directly transferred to the weight memory 1102 through the direct memory access controller (DMAC) 1105. The input data is also transferred to the unified memory 1106 through the DMAC.

BIU stands for Bus Interface Unit, that is, the bus interface unit 1110, which is used for the interaction between AXI bus and DMAC and instruction fetch buffer (IFB) 1109.

The bus interface unit 1110 (BIU for short) is used for the instruction fetch memory 1109 to obtain instructions from the external memory, and is also used for the storage unit access controller 1105 to obtain the original data of the input matrix A or the weight matrix B from the external memory.

DMAC is mainly used to transfer input data in the external memory DDR to the unified memory 1106 or to transfer weight data to the weight memory 1102 or to transfer input data to the input memory 1101.

The vector calculation unit 1107 includes multiple operation processing units, and when necessary, further processes the output of the operation circuit 1103, such as vector multiplication, vector addition, exponential operation, logarithmic operation, size comparison, etc. It is mainly used for non-convolutional/fully connected layer network calculations in neural networks, such as Batch Normalization, pixel-level summation, upsampling of feature planes, etc.

In some implementations, the vector calculation unit 1107 can store the processed output vector to the unified memory 1106. For example, the vector calculation unit 1107 can apply a linear function; or a nonlinear function to the output of the operation circuit 1103, such as linear interpolation of the feature plane extracted by the convolution layer, and then, for example, a vector of accumulated values to generate an activation value. In some implementations, the vector calculation unit 1107 generates a normalized value, a pixel-level summed value, or both. In some implementations, the processed output vector can be used as an activation input to the operation circuit 1103, for example, for use in a subsequent layer in a neural network.

An instruction fetch buffer 1109 connected to the controller 1104 is used to store instructions used by the controller 1104;

Unified memory 1106, input memory 1101, weight memory 1102 and instruction fetch memory 1109 are all on-chip memories. External memories are private to the NPU hardware architecture.

The processor mentioned in any of the above places may be a general-purpose central processing unit, a microprocessor, an ASIC, or one or more integrated circuits for controlling the execution of the above programs.

It should also be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the device embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.

Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a magnetic disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the methods described in each embodiment of the present application.

In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, Computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, training device or data center to another website, computer, training device or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device, data center, etc. that includes one or more available media integrated. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid state drive (SSD)).

Claims (13)

A data processing method, characterized in that the method comprises: Obtaining an image and an object detection request, wherein the object detection request includes a natural language description of an object to be detected in the image; Processing the image by an image encoder to obtain a feature representation of each of a plurality of image regions, each image region corresponding to a candidate detection frame; The object detection request and the plurality of feature representations are processed by a language model, and the region and category of the object to be detected are determined from the plurality of image regions. The method according to claim 1, characterized in that the step of processing the object detection request and the plurality of feature representations through a language model, and determining the region where the object to be detected is located from the plurality of image regions, comprises: Processing the object detection request and each of the feature representations in parallel through a language model to obtain a detection result for each of the image regions; According to the detection result, the area where the object to be detected is located is determined from the multiple image areas. The method according to claim 1 or 2, characterized in that before processing the image by the image encoder, the method further comprises: Acquire a training sample, wherein the training sample includes an image sample and a corresponding text sample, wherein the text sample is a text description corresponding to the image sample; Processing the image sample through an image encoder to obtain a first processing result; Processing the text sample through a text encoder to obtain a second processing result; According to the first processing result and the second processing result, the image encoder and the text encoder are updated through comparative learning. The method according to claim 3, characterized in that the first processing result includes multiple detection boxes and a category of each detection box; the method further comprises: The image encoder and the text encoder are updated according to the first processing result and the corresponding true value. The method according to any one of claims 1 to 4 is characterized in that the language model includes multiple network layers, at least one of the multiple network layers includes a cross-attention layer; the cross-attention layer is used to perform attention interaction between image features and text features. A data processing device, characterized in that the device comprises: An acquisition module, configured to acquire an image and an object detection request, wherein the object detection request includes a natural language description of an object to be detected in the image; A processing module, configured to process the image through an image encoder to obtain a feature representation of each of a plurality of image regions, each image region corresponding to a candidate detection frame; The object detection request and the plurality of feature representations are processed by a language model, and the region and category of the object to be detected are determined from the plurality of image regions. The device according to claim 6, characterized in that the processing module is specifically used to: Processing the object detection request and each of the feature representations in parallel through a language model to obtain a detection result for each of the image regions; According to the detection result, the area where the object to be detected is located is determined from the multiple image areas. The device according to claim 6 or 7, characterized in that before the image is processed by the image encoder, the acquisition module is further used to: Acquire a training sample, wherein the training sample includes an image sample and a corresponding text sample, wherein the text sample is a text description corresponding to the image sample; The processing module is further used to: process the image sample through an image encoder to obtain a first processing result; Processing the text sample through a text encoder to obtain a second processing result; According to the first processing result and the second processing result, the image encoder and the text encoder are updated through comparative learning. The device according to claim 8, wherein the first processing result includes a plurality of detection frames and a category of each detection frame; and the processing module is further configured to: The image encoder and the text encoder are updated according to the first processing result and the corresponding true value. The device according to any one of claims 6 to 9 is characterized in that the language model includes multiple network layers, at least one of the multiple network layers includes a cross-attention layer; the cross-attention layer is used to perform attention interaction between image features and text features. A computer storage medium, characterized in that the computer storage medium stores one or more instructions, which, when executed by one or more computers, enable the one or more computers to perform the operations of the method described in any one of claims 1 to 5. A computer program product, characterized in that it comprises computer-readable instructions, and when the computer-readable instructions are executed on a computer device, the computer device is caused to execute the method according to any one of claims 1 to 5. A system comprises at least one processor and at least one memory; the processor and the memory are connected via a communication bus and communicate with each other; The at least one memory is used to store code; The at least one processor is configured to execute the code to perform the method according to any one of claims 1 to 5.