CN111612105A - Image prediction method, apparatus, computer equipment and storage medium - Google Patents

Abstract

The present application relates to an image prediction method, apparatus, computer equipment and storage medium. The method includes: acquiring a sequence of images to be tested; the sequence of images to be tested includes at least two images to be tested; inputting the sequence of images to be tested into a preset image prediction model to obtain a predicted image sequence; wherein, the image prediction model is based on the prediction method. The preset loss function is obtained by training, and the preset loss function is obtained by adjusting the weight of each sub-function according to the image loss value between the standard image sequence and the sample prediction image sequence. By using this method, a relatively accurate loss function can be obtained by adjusting the weights of the sub-functions of the loss function of the image prediction model, and the preset image prediction model can be accurately trained, which improves the accuracy of the obtained image prediction model, and can be used according to The obtained image prediction model with higher accuracy can obtain an accurately predicted image sequence, thereby improving the consistency of the obtained predicted image sequence.

Description

Image prediction method, apparatus, computer equipment and storage medium

technical field

The present application relates to the technical field of image prediction, and in particular, to an image prediction method, apparatus, computer equipment and storage medium.

Background technique

With the development of machine learning technology, great progress has been made in the application of machine learning in all walks of life, among which deep learning based on big data is a new direction in the field of machine learning.

The recurrent neural network in deep learning is a type of recurrent neural network that takes sequence data as input, performs recursion in the evolution direction of the sequence, and connects all nodes in a chain. Recurrent neural network has memory, parameter sharing and Turing completeness, so it has certain advantages in learning the nonlinear characteristics of sequences, and has a very bright performance in some image sequence prediction tasks.

However, when dealing with these sequence generation tasks, the recurrent neural network has the problem of poor consistency of the generated image sequences.

SUMMARY OF THE INVENTION

Based on this, it is necessary to provide an image prediction method, apparatus, computer equipment and storage medium that can improve the consistency of the generated image sequence in response to the above technical problems.

An image prediction method, the method comprising:

acquiring a sequence of images to be tested; the sequence of images to be tested includes at least two images to be tested;

Input the image sequence to be tested into a preset image prediction model to obtain a predicted image sequence; wherein, the image prediction model is obtained by training according to a preset loss function, and the preset loss function is based on a standard image sequence The image loss value between the sample prediction image sequence and the weight of each sub-function is adjusted.

In one embodiment, the training method of the image prediction model includes:

Input the sample image sequence into the preset image prediction model to be trained to obtain the sample prediction image sequence;

Adjust the weight of each sub-function in the loss function of the image prediction model to be trained according to the image loss value between the standard image sequence and the sample prediction image sequence;

The image prediction model to be trained is trained according to the adjusted value of the loss function to obtain the image prediction model.

In one embodiment, the image prediction model to be trained is trained according to the value of the adjusted loss function to obtain the image prediction model, including:

According to the value of the adjusted loss function, the parameters of the image prediction model to be trained are adjusted to obtain a new image prediction model to be trained;

Taking the new image prediction model to be trained as the preset image prediction model to be trained, and returning to the step of inputting the sample image sequence into the preset image prediction model to be trained to obtain the sample prediction image sequence, until the preset image sequence is obtained. The image prediction model is obtained when the set convergence conditions are met.

In one embodiment, the preset convergence condition includes that the value of the adjusted loss function reaches a stable value or reaches a preset number of iterations.

In one embodiment, according to the image loss value between the standard image sequence and the sample prediction image sequence, adjusting the weight of each sub-function in the loss function of the image prediction model to be trained includes:

Obtain the image loss value between each sample predicted image and the corresponding standard predicted image in the sample predicted image sequence;

According to the image loss value between the predicted image of each sample and the corresponding standard predicted image, the weight of the corresponding sub-function in the loss function is adjusted.

In one of the embodiments, the larger the image loss value is, the larger the weight of the corresponding sub-function is.

A training method for an image prediction model, the method comprising:

Input the sample image sequence into the preset image prediction model to be trained to obtain the sample prediction image sequence;

Adjust the weight of each sub-function in the loss function of the image prediction model to be trained according to the image loss value between the standard image sequence and the sample prediction image sequence;

The image prediction model to be trained is trained according to the value of the adjusted loss function to obtain the image prediction model.

An image prediction apparatus, the apparatus includes:

a first acquisition module, configured to acquire a sequence of images to be tested; the sequence of images to be tested includes at least two images to be tested;

a prediction module, configured to input the image sequence to be tested into a preset image prediction model to obtain a predicted image sequence; wherein, the image prediction model is obtained by training according to a preset loss function, and the preset loss function It is obtained by adjusting the weight of each sub-function to predict the image loss value between the standard image sequence and the sample image sequence.

A computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

acquiring a sequence of images to be tested; the sequence of images to be tested includes at least two images to be tested;

Input the image sequence to be tested into a preset image prediction model to obtain a predicted image sequence; wherein, the image prediction model is obtained by training according to a preset loss function, and the preset loss function is based on a standard image sequence The image loss value between the sample prediction image sequence and the weight of each sub-function is adjusted.

A computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

acquiring a sequence of images to be tested; the sequence of images to be tested includes at least two images to be tested;

Input the image sequence to be tested into a preset image prediction model to obtain a predicted image sequence; wherein, the image prediction model is obtained by training according to a preset loss function, and the preset loss function is based on a standard image sequence The image loss value between the sample prediction image sequence and the weight of each sub-function is adjusted.

The above image prediction method, device, computer equipment and storage medium can obtain a relatively accurate loss function by adjusting the weight of each sub-function of the loss function of the image prediction model, so that the preset image prediction model can be performed according to the obtained loss function. Accurate training improves the accuracy of the obtained image prediction model, and then an accurately predicted image sequence can be obtained according to the obtained image prediction model with higher accuracy, thereby improving the consistency of the obtained predicted image sequence.

Description of drawings

1 is a schematic diagram of the internal structure of a computer device provided by an embodiment;

2 is a schematic flowchart of an image prediction method provided by an embodiment;

3 is a schematic flowchart of an image prediction method provided by another embodiment;

4 is a schematic flowchart of a training method for an image prediction model provided by an embodiment;

5 is a schematic flowchart of an image prediction method provided by an embodiment;

6 is a schematic structural diagram of an image prediction apparatus provided by an embodiment;

FIG. 7 is a schematic structural diagram of an apparatus for training an image prediction model according to an embodiment.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application will be described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, but not to limit the present application.

The image prediction method provided in the embodiment of the present application can be applied to the computer device as shown in FIG. 1 . The computer device includes a processor and a memory connected through a system bus, where a computer program is stored in the memory, and the processor can execute the steps of the following method embodiments when the processor executes the computer program. Optionally, the computer equipment may further include a network interface, a display screen and an input device. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium storing an operating system and a computer program, and an internal memory. The internal memory provides an environment for the execution of the operating system and computer programs in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. Optionally, the computer device may be a server, a personal computer, a personal digital assistant, or other terminal devices, such as a tablet computer, a mobile phone, etc., or a cloud or a remote server, which is implemented in this application. The example does not limit the specific form of the computer equipment.

In one embodiment, as shown in FIG. 2, an image prediction method is provided, and the method is applied to the computer device in FIG. 1 as an example for description, including the following steps:

S201 , acquiring a sequence of images to be tested; the sequence of images to be tested includes at least two images to be tested.

Wherein, the image sequence to be tested is composed of a group of continuous images to be tested, including at least two images to be tested. Specifically, the computer device acquires a sequence of images to be tested including at least two images to be tested. Optionally, the computer equipment can perform frame extraction on a video to obtain the image sequence to be tested, or a set of continuous images to be tested can be captured by a camera device to obtain the image sequence to be tested.

S202: Input the image sequence to be tested into a preset image prediction model to obtain a predicted image sequence. The image prediction model is obtained by training according to a preset loss function, and the preset loss function is obtained by adjusting the weight of each sub-function according to the image loss value between the standard image sequence and the sample predicted image sequence.

式中，L表示该图像预测模型的预设的损失函数，L_i表示该损失函数的各子函数，W_i表示该损失函数的各子函数的权重，计算机设备根据得到的预设的损失函数的值对初始图像预测模型进行训练，从而得到该图像预测模型。Specifically, the computer device inputs the acquired image sequence to be tested into a preset image prediction model to obtain a predicted image sequence. The image prediction model is obtained by training according to a preset loss function, and the preset loss function is the image loss value between the standard image sequence and the sample prediction image sequence, after adjusting the weight of each sub-function of the loss function. owned. Optionally, the image prediction model can be a recurrent neural network model. Exemplarily, assuming that the image sequence to be tested is (K1, K2,..., K10) and the predicted image sequence is (T1, T2,..., T10), then the preset loss function of the image prediction model is

In the formula, _L represents the preset loss function of the image prediction model, Li represents each sub-function of the loss function, _Wi represents the weight of each sub-function of the loss function, and the computer equipment obtains the preset loss function according to The value of , trains the initial image prediction model to obtain the image prediction model.

In this embodiment, by adjusting the weights of the sub-functions of the loss function of the image prediction model, a relatively accurate loss function can be obtained, so that the preset image prediction model can be accurately trained according to the obtained loss function, and the result is improved. The accuracy of the obtained image prediction model can then be used to obtain an accurately predicted image sequence according to the obtained image prediction model with higher accuracy, thereby improving the consistency of the obtained predicted image sequence.

In one embodiment, as shown in FIG. 3 , on the basis of the foregoing embodiment, as an optional implementation manner, the training method of the foregoing image prediction model includes:

S301, input the sample image sequence into a preset image prediction model to be trained to obtain the sample prediction image sequence.

Specifically, the computer device inputs the sample image sequence into a preset image prediction model to be trained, and obtains the sample prediction image sequence. Optionally, the preset image prediction model to be trained may be a recurrent neural network model. Exemplarily, assuming that the sample image sequence is represented as the first 10 frames of images as (S1, S2, ..., S10), the next 10 frames of images to be predicted can be represented as (H1, H2, ..., H10), then the computer The device can input the sample image sequence (S1, S2, ..., S10) into the preset image prediction model to be trained, and obtain the sample predicted image sequence (H1, H2, ..., H10).

S302, according to the image loss value between the standard image sequence and the sample prediction image sequence, adjust the weight of each sub-function in the loss function of the image prediction model to be trained.

Specifically, the computer device adjusts the weight of each sub-function in the loss function of the image prediction model to be trained according to the image loss value between the standard image sequence and the sample prediction image sequence. Optionally, the computer device can obtain the image loss value between each sample predicted image in the sample predicted image sequence and the corresponding standard predicted image, and based on the difference between each sample predicted image in the sample predicted image sequence and the corresponding standard predicted image. , adjust the weight of the corresponding sub-function in the loss function of the image prediction model to be trained.

S303 , train the image prediction model to be trained according to the value of the adjusted loss function to obtain an image prediction model.

Specifically, the computer device trains the image prediction model to be trained according to the value of the adjusted loss function of the image prediction model to be trained, to obtain the above image prediction model. Optionally, the computer device can adjust the parameters of the image prediction model to be trained according to the value of the adjusted loss function to obtain a new image prediction model to be trained, and then use the new image prediction model to be trained as the preset to be trained. The image prediction model is returned, and the step of inputting the sample image sequence into the preset image prediction model to be trained to obtain the sample prediction image sequence is performed, and the above image prediction model is obtained when the preset convergence condition is reached. Optionally, the preset convergence condition may include that the value of the loss function after the adjustment of the image prediction model to be trained reaches a stable value or reaches a preset number of iterations, for example, the preset number of iterations may be 10 times, or 20 times. inferior. Exemplarily, the computer device can adjust the parameters of the image prediction model to be trained according to the value of the loss function adjusted by the image prediction model to be trained, obtain a new image prediction model to be trained, and then input the sample image sequence into the new image prediction model to be trained. Train the image prediction model, obtain a new sample prediction image sequence, predict the image loss value between the standard image sequence and the new sample image sequence, adjust the weight of each sub-function in the loss function of the image prediction model to be trained, and adjust according to the The new image prediction model to be trained is trained with the value of the lost loss function, and this step is repeated until the image prediction model is obtained when a preset convergence condition is reached.

In this embodiment, the computer device inputs the sample image sequence into the preset image prediction model to be trained, and can obtain the sample prediction image sequence, so that the image loss value between the standard image sequence and the sample prediction image sequence can be accurately obtained, and then According to the image loss value between the standard image sequence and the sample prediction image sequence, the weight of each sub-function in the loss function of the image prediction model to be trained can be accurately adjusted, and the image prediction model to be trained can be adjusted according to the value of the adjusted loss function. Accurate training improves the accuracy of the obtained image prediction model, and then an accurately predicted image sequence can be obtained according to the obtained image prediction model with higher accuracy, thereby improving the consistency of the obtained predicted image sequence.

In the above scenario of adjusting the weight of the corresponding sub-function in the loss function according to the image loss value between each sample predicted image and the corresponding standard predicted image, on the basis of the above embodiment, as an optional implementation , the greater the image loss value, the greater the weight of the corresponding sub-function.

Specifically, according to the obtained image loss value between each sample predicted image and the corresponding standard predicted image, the computer device increases the weight of the sub-function corresponding to the larger image loss value, and the smaller image loss value corresponds to the weight of the sub-function. The weights of the sub-functions are adjusted down. Optionally, the computer device may adjust the weight of the sub-function corresponding to the larger image loss value to be greater than 1, and adjust the weight of the sub-function corresponding to the smaller image loss value to be smaller than 1.

In this embodiment, the larger the image loss value is, the larger the weight of the corresponding sub-function is, so that the value of the loss function of the image prediction model to be trained can be adaptively adjusted according to the weight of the adjusted sub-function. The value of the loss function of the image prediction model is adjusted adaptively to the parameters of the image prediction model to be trained, so that the image prediction model to be trained pays more attention to the places where the image loss value is relatively large, so that the image prediction model to be trained after adjusting the parameters is adjusted. The sample prediction image sequence can be obtained accurately.

In one embodiment, as shown in FIG. 4 , a training method for an image prediction model is provided, and the method is applied to the computer device in FIG. 1 as an example to illustrate, including the following steps:

S401: Input the sample image sequence into a preset image prediction model to be trained to obtain a sample prediction image sequence.

Specifically, the computer device inputs the sample image sequence into a preset image prediction model to be trained, and obtains the sample prediction image sequence. Optionally, the preset image prediction model to be trained may be a recurrent neural network model. Exemplarily, assuming that the sample image sequence is represented as the first 10 frames of images as (S1, S2, ..., S10), the next 10 frames of images to be predicted can be represented as (H1, H2, ..., H10), then the computer The device can input the sample image sequence (S1, S2, ..., S10) into the preset image prediction model to be trained, and obtain the sample predicted image sequence (H1, H2, ..., H10).

S402, according to the image loss value between the standard image sequence and the sample prediction image sequence, adjust the weight of each sub-function in the loss function of the image prediction model to be trained.

Specifically, the computer device adjusts the weight of each sub-function in the loss function of the image prediction model to be trained according to the image loss value between the standard image sequence and the sample prediction image sequence. Optionally, the computer device can obtain the image loss value between each sample predicted image in the sample predicted image sequence and the corresponding standard predicted image, and based on the difference between each sample predicted image in the sample predicted image sequence and the corresponding standard predicted image. , adjust the weight of the corresponding sub-function in the loss function of the image prediction model to be trained. Optionally, the larger the image loss value, the larger the weight of the corresponding sub-function.

S403 , train the image prediction model to be trained according to the value of the adjusted loss function to obtain an image prediction model.

Specifically, the computer device trains the image prediction model to be trained according to the value of the adjusted loss function, until the value of the adjusted loss function reaches a stable value or reaches a preset number of iterations to obtain the image prediction model. Optionally, the computer device can adjust the parameters of the image prediction model to be trained according to the value of the loss function adjusted by the image prediction model to be trained, obtain a new image prediction model to be trained, and then input the sample image sequence into the new image prediction model to be trained. Train the image prediction model, obtain a new sample prediction image sequence, predict the image loss value between the standard image sequence and the new sample image sequence, adjust the weight of each sub-function in the loss function of the image prediction model to be trained, and adjust according to the The new image prediction model to be trained is trained with the value of the lost loss function, and this step is repeated until the value of the adjusted loss function reaches a stable value or reaches a preset number of iterations to obtain the image prediction model.

In this embodiment, the computer device inputs the sample image sequence into the preset image prediction model to be trained, and can obtain the sample prediction image sequence, so that the image loss value between the standard image sequence and the sample prediction image sequence can be accurately obtained, and then According to the image loss value between the standard image sequence and the sample prediction image sequence, the weight of each sub-function in the loss function of the image prediction model to be trained can be accurately adjusted, and the image prediction model to be trained can be adjusted according to the value of the adjusted loss function. Accurate training improves the accuracy of the obtained image prediction model, and then an accurately predicted image sequence can be obtained according to the obtained image prediction model with higher accuracy, thereby improving the consistency of the obtained predicted image sequence.

In order to facilitate the understanding of those skilled in the art, the following describes the image prediction method provided by the present application in detail. As shown in FIG. 5 , the method may include:

S501, input the sample image sequence into a preset image prediction model to be trained to obtain the sample prediction image sequence;

S502, obtaining the image loss value between each sample predicted image and the corresponding standard predicted image in the sample predicted image sequence;

S503, according to the image loss value between each sample predicted image and the corresponding standard predicted image, adjust the weight of the corresponding sub-function in the loss function;

S504, according to the value of the adjusted loss function, adjust the parameters of the image prediction model to be trained to obtain a new image prediction model to be trained;

S505, take the new image prediction model to be trained as the preset image prediction model to be trained, and return to the step of inputting the sample image sequence into the preset image prediction model to be trained to obtain the sample prediction image sequence, until the preset image prediction model is reached The image prediction model is obtained when the convergence conditions are met; the preset convergence conditions include that the value of the adjusted loss function reaches a stable value or reaches a preset number of iterations;

S506, obtain a sequence of images to be tested; the sequence of images to be tested includes at least two images to be tested;

S507: Input the image sequence to be tested into the image prediction model to obtain a predicted image sequence.

It should be noted that, for the descriptions in the foregoing S501-S509, reference may be made to the relevant descriptions in the foregoing embodiments, and the effects thereof are similar, and details are not described herein again in this embodiment.

It should be understood that although the steps in the flowcharts of FIGS. 2-5 are shown in sequence according to the arrows, these steps are not necessarily executed in the sequence shown by the arrows. Unless explicitly stated herein, the execution of these steps is not strictly limited to the order, and these steps may be performed in other orders. Moreover, at least a part of the steps in FIGS. 2-5 may include multiple steps or multiple stages. These steps or stages are not necessarily executed and completed at the same time, but may be executed at different times. The execution of these steps or stages The order is also not necessarily sequential, but may be performed alternately or alternately with other steps or at least a portion of the steps or phases within the other steps.

In one embodiment, as shown in FIG. 6, an image prediction apparatus is provided, including: a first acquisition module and a prediction module, wherein:

The first acquisition module is used to acquire a sequence of images to be tested; the sequence of images to be tested includes at least two images to be tested.

a prediction module, configured to input the image sequence to be tested into a preset image prediction model to obtain a predicted image sequence; wherein, the image prediction model is obtained by training according to a preset loss function, and the preset loss function It is obtained by adjusting the weight of each sub-function to predict the image loss value between the standard image sequence and the sample image sequence.

The image prediction apparatus provided in this embodiment can execute the above method embodiments, and the implementation principles and technical effects thereof are similar, and details are not described herein again.

On the basis of the above-mentioned embodiment, optionally, the above-mentioned apparatus further includes: a second acquisition module, a first adjustment module and a first training module, wherein:

The second acquisition module is used to input the sample image sequence into the preset image prediction model to be trained to obtain the sample prediction image sequence;

a first adjustment module, configured to adjust the weight of each sub-function in the loss function of the image prediction model to be trained according to the image loss value between the standard image sequence and the sample prediction image sequence;

The first training module is configured to train the image prediction model to be trained according to the value of the adjusted loss function to obtain the image prediction model.

The image prediction apparatus provided in this embodiment can execute the above method embodiments, and the implementation principles and technical effects thereof are similar, and details are not described herein again.

On the basis of the above-mentioned embodiment, optionally, the above-mentioned first training module includes: a first adjustment unit and a training unit, wherein:

a first adjustment unit, configured to adjust the parameters of the image prediction model to be trained according to the value of the adjusted loss function to obtain a new image prediction model to be trained;

The training unit is configured to use the new image prediction model to be trained as the preset image prediction model to be trained, and return to execute the inputting the sample image sequence into the preset image prediction model to be trained, and obtain a sample image prediction model of the image sequence. step, until a preset convergence condition is reached, the image prediction model is obtained.

Optionally, the preset convergence condition includes that the value of the adjusted loss function reaches a stable value or reaches a preset number of iterations.

The image prediction apparatus provided in this embodiment can execute the above method embodiments, and the implementation principles and technical effects thereof are similar, and details are not described herein again.

On the basis of the above-mentioned embodiment, optionally, the above-mentioned first adjustment module includes: an acquisition unit and a second adjustment unit, wherein:

an acquisition unit for acquiring the image loss value between each sample predicted image and the corresponding standard predicted image in the sample predicted image sequence;

The second adjustment unit is configured to adjust the weight of the corresponding sub-function in the loss function according to the image loss value between the predicted image of each sample and the corresponding standard predicted image.

Optionally, the larger the image loss value is, the larger the weight of the corresponding sub-function is.

The image prediction apparatus provided in this embodiment can execute the above method embodiments, and the implementation principles and technical effects thereof are similar, and details are not described herein again.

For the specific definition of the image prediction apparatus, reference may be made to the above definition of the image prediction method, which will not be repeated here. Each module in the above-mentioned image prediction apparatus may be implemented in whole or in part by software, hardware and combinations thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

In one embodiment, as shown in FIG. 7 , an apparatus for training an image prediction model is provided, comprising: a third acquisition module, a second adjustment module and a second training module, wherein:

The third acquisition module is used to input the sample image sequence into the preset image prediction model to be trained to obtain the sample prediction image sequence;

The second adjustment module is configured to adjust the weight of each sub-function in the loss function of the image prediction model to be trained according to the image loss value between the standard image sequence and the sample prediction image sequence;

The second training module is configured to train the image prediction model to be trained according to the value of the adjusted loss function to obtain the image prediction model.

The apparatus for training an image prediction model provided in this embodiment can execute the above method embodiments, and the implementation principles and technical effects thereof are similar, which will not be repeated here.

For the specific definition of the apparatus for training the image prediction model, reference may be made to the definition of the training method for the image prediction model above, which will not be repeated here. Each module in the above image prediction model training device can be implemented in whole or in part by software, hardware and combinations thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

In one embodiment, a computer device is provided, including a memory and a processor, a computer program is stored in the memory, and the processor implements the following steps when executing the computer program:

acquiring a sequence of images to be tested; the sequence of images to be tested includes at least two images to be tested;

Input the image sequence to be tested into a preset image prediction model to obtain a predicted image sequence; wherein, the image prediction model is obtained by training according to a preset loss function, and the preset loss function is based on a standard image sequence The image loss value between the sample prediction image sequence and the weight of each sub-function is adjusted.

The implementation principles and technical effects of the computer equipment provided by the above embodiments are similar to those of the above method embodiments, and details are not described herein again.

In one embodiment, a computer-readable storage medium is provided on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

acquiring a sequence of images to be tested; the sequence of images to be tested includes at least two images to be tested;

Input the image sequence to be tested into a preset image prediction model to obtain a predicted image sequence; wherein, the image prediction model is obtained by training according to a preset loss function, and the preset loss function is based on a standard image sequence The image loss value between the sample prediction image sequence and the weight of each sub-function is adjusted.

The implementation principles and technical effects of the computer-readable storage medium provided by the foregoing embodiments are similar to those of the foregoing method embodiments, and details are not described herein again.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage In the medium, when the computer program is executed, it may include the processes of the above-mentioned method embodiments. Wherein, any reference to memory, storage, database or other media used in the various embodiments provided in this application may include at least one of non-volatile and volatile memory. The non-volatile memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, and the like. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, the RAM may be in various forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM).

The technical features of the above embodiments can be combined arbitrarily. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features It is considered to be the range described in this specification.

The above-mentioned embodiments only represent several embodiments of the present application, and the descriptions thereof are specific and detailed, but should not be construed as a limitation on the scope of the invention patent. It should be pointed out that for those skilled in the art, without departing from the concept of the present application, several modifications and improvements can be made, which all belong to the protection scope of the present application. Therefore, the scope of protection of the patent of the present application shall be subject to the appended claims.

Claims (10)

1. A method of image prediction, the method comprising:

acquiring an image sequence to be detected; the image sequence to be detected comprises at least two images to be detected;

inputting the image sequence to be detected into a preset image prediction model to obtain a predicted image sequence; the image prediction model is obtained by training according to a preset loss function, and the preset loss function is obtained by adjusting the weight of each sub-function according to the image loss value between a standard image sequence and a sample prediction image sequence.

2. The method of claim 1, wherein the training method of the image prediction model comprises:

inputting the sample image sequence into a preset image prediction model to be trained to obtain a sample prediction image sequence;

adjusting the weight of each sub-function in the loss function of the image prediction model to be trained according to the image loss value between the standard image sequence and the sample prediction image sequence;

and training the image prediction model to be trained according to the adjusted value of the loss function to obtain the image prediction model.

3. The method according to claim 2, wherein the training the image prediction model to be trained according to the adjusted value of the loss function to obtain the image prediction model comprises:

adjusting parameters of the image prediction model to be trained according to the adjusted value of the loss function to obtain a new image prediction model to be trained;

and taking the new image prediction model to be trained as a preset image prediction model to be trained, and returning to the step of inputting the sample image sequence into the preset image prediction model to be trained to obtain a sample predicted image sequence until a preset convergence condition is reached to obtain the image prediction model.

4. The method of claim 3, wherein the preset convergence condition comprises the adjusted loss function reaching a stable value or reaching a preset number of iterations.

5. The method according to claim 2, wherein the adjusting the weight of each sub-function in the loss function of the image prediction model to be trained according to the image loss value between the standard image sequence and the sample prediction image sequence comprises:

acquiring an image loss value between each sample predicted image in the sample predicted image sequence and the corresponding standard predicted image;

and adjusting the weight of the corresponding sub-function in the loss function according to the image loss value between each sample prediction image and the corresponding standard prediction image.

6. The method of claim 5, wherein the greater the image loss value, the greater the weight of the corresponding sub-function.

7. A method for training an image prediction model, the method comprising:

inputting the sample image sequence into a preset image prediction model to be trained to obtain a sample prediction image sequence;

adjusting the weight of each sub-function in the loss function of the image prediction model to be trained according to the image loss value between the standard image sequence and the sample prediction image sequence;

and training the image prediction model to be trained according to the adjusted value of the loss function to obtain the image prediction model.

8. An image prediction apparatus, characterized in that the apparatus comprises:

the first acquisition module is used for acquiring an image sequence to be detected; the image sequence to be detected comprises at least two images to be detected;

the prediction module is used for inputting the image sequence to be detected into a preset image prediction model to obtain a predicted image sequence; the image prediction model is obtained by training according to a preset loss function, and the preset loss function is obtained by adjusting the weight of each sub-function according to the image loss value between a standard image sequence and a sample prediction image sequence.

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 7.

10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.

Images