CN115423770A - Pedestrian re-identification data discrimination method based on image quality discrimination model - Google Patents
Pedestrian re-identification data discrimination method based on image quality discrimination model Download PDFInfo
- Publication number
- CN115423770A CN115423770A CN202211059224.0A CN202211059224A CN115423770A CN 115423770 A CN115423770 A CN 115423770A CN 202211059224 A CN202211059224 A CN 202211059224A CN 115423770 A CN115423770 A CN 115423770A
- Authority
- CN
- China
- Prior art keywords
- image
- quality
- pedestrian
- decoder
- classifier
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/103—Static body considered as a whole, e.g. static pedestrian or occupant recognition
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Multimedia (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a pedestrian re-identification data discrimination method based on an image quality discrimination model, which comprises the following steps of 1, determining an image quality evaluation standard; step 2, acquiring a pedestrian image, labeling the pedestrian and image quality according to the standard in the step 1 to form a sample set, and distinguishing a high-quality image and a low-quality image for subsequent training; step 3, preprocessing the collected pedestrian image, and step 4, training an image quality discrimination model based on a reversible decoder by using the high-quality image data set obtained in the step 3; and 5, accessing the image quality discrimination model based on the reversible decoder into the device. Through carrying out quality evaluation to pedestrian re-identification data, promote the rate of accuracy of re-identification, retrain the data bulk, reduce the calculation power demand of re-identification, effectively promote efficiency.
Description
Technical Field
The invention relates to the field of computer vision, intelligent security and smart city research, in particular to a pedestrian re-identification data discrimination method based on an image quality discrimination model.
Background
With the continuous popularization of artificial intelligence in China, the pedestrian re-identification technology is developed and applied in various fields, and the management efficiency of intelligent security and smart city is greatly improved. However, although the deep learning pedestrian re-recognition model is excellent in the task of re-recognition of pedestrians, it is limited by various conditions, and the collected images have common image quality problems such as blur, noise and low resolution, which affect the recognition effect, so that the quality judgment of the images becomes one of important links in the task of re-recognition of pedestrians.
The deep learning pedestrian re-identification model can solve the problems of identification and retrieval of pedestrians in a cross-camera and cross-scene mode. The quality requirement of current pedestrian heavy identification model to the pedestrian image is very high, however in-process of in-service use, the pedestrian under the surveillance camera head is all less, and resolution ratio is also very low, and the problem that the surveillance image of some places was sheltered from with there are a lot of pedestrians is no matter be the sheltering from of pedestrian and building, or the sheltering from between pedestrian and the pedestrian, all can seriously influence the rate of recognition.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention provides a pedestrian re-identification data discrimination method based on an image quality discrimination model, which is used for carrying out quality evaluation on pedestrian re-identification data, improving the accuracy of re-identification, constraining the data quantity, reducing the computational power requirement of re-identification and effectively improving the efficiency. The technical scheme is as follows:
the invention provides a pedestrian re-identification data discrimination method based on an image quality discrimination model, which mainly comprises the following steps:
step 1, determining an image quality evaluation standard;
and (3) formulating a pedestrian image quality evaluation standard according to the requirement of the pedestrian re-identification module, setting a threshold value, and dividing the image into a high-quality image and a low-quality image.
Step 2, acquiring a pedestrian image, labeling the pedestrian and image quality according to the standard in the step 1 to form a sample set, and distinguishing a high-quality image and a low-quality image for subsequent training;
step 3, preprocessing the collected pedestrian image, specifically: the mean and variance of the image data in the sample set are calculated and the pixel values of the image are normalized to [0,1] and then scaled to the appropriate size, resulting in a high quality image data set and a low quality image data set.
Step 4, training an image quality discrimination model based on a reversible decoder by using the high-quality image data set obtained in the step 3;
reversible decoder training: using a reversible depth model i-RevNet as a decoder and the inverse of i-RevNet as an encoder, inputting the high-quality image data set in the step 3 into the encoder, connecting an output layer of the encoder with an input layer of the decoder, restoring the characteristics into an image by the decoder, calculating the performance of the decoder through a formula 1, and reducing L through multiple rounds of training softmax So that the decoder can learn the characteristics of a high-quality graph as much as possible to obtain the decoder for training the classifier;
equation 1:
wherein theta and omega are parameters required to be learned by the model, and high-quality diagram features
The characteristic x of the low-quality image is as a function of
y ∈ γ = {0,1, \8230;, c } represents a set of quantity categories in c,
is a plurality of
y, and H (a) is the feature extracted by the feature extractor.
Training a quality discrimination classifier: fixing the i-RevNet parameters of the decoder, selecting ResNet as an image quality classifier, inputting the high-quality image data set and the low-quality image data set in the step 3 into the classifier, connecting an output layer of the classifier with an input layer of the decoder, calculating the performance of the encoder and the classifier through a formula 2 for judging the training effect, obtaining the trained classifier after multiple rounds of training, and outputting the classifier as an image quality score.
Equation 2:
wherein
Is a parameter that needs to be trained at the current stage,
parameters fixed for the decoder;
the classifier is extracted as an image quality discrimination model based on a reversible decoder.
Step 5, accessing the image quality discrimination model based on the reversible decoder into the device;
and sending the image to be identified into an image quality discrimination model based on a reversible decoder to obtain a quality discrimination score. A pedestrian re-identification module calculates a final pedestrian identification result in combination with the quality score.
Preferably, the evaluation criteria for the image quality of the pedestrian in step 1 include one or more of the definition of the image, the presence of noise, the complete definition of the pedestrian, the complexity of the background, the image resolution of the pedestrian, the shooting angle and the light source.
Preferably, in step 2, in order to ensure the integrity of the sample set in the time dimension, when the images of the pedestrians are collected, a certain amount of images are respectively intercepted from the videos in four time periods of morning, noon, afternoon and evening.
Further, step 2, the step of collecting the pedestrian image further comprises collecting the pedestrian image output by the pedestrian detection module.
Compared with the prior art, one of the technical schemes has the following beneficial effects: before pedestrian re-identification, the image quality is evaluated by using an image quality discrimination model based on a reversible decoder, then a quality evaluation score is determined according to the requirement of a pedestrian re-identification module, and whether identification is carried out or not is determined so as to improve the identification efficiency and accuracy of the device. By constraining the data volume, the computational power requirement of re-identification is reduced, and the efficiency is effectively improved. Meanwhile, the quality discrimination model trained by the method has better performance, can integrate more factors influencing the quality, and optimizes the effect of pedestrian re-identification.
Drawings
Fig. 1 is a flowchart of a pedestrian re-identification data discrimination method based on an image quality discrimination model according to an embodiment of the present disclosure.
Fig. 2 is a schematic diagram of an image quality discrimination model according to an embodiment of the disclosure.
Detailed Description
In order to clarify the technical solution and the working principle of the present invention, the following detailed description will be made on the embodiments of the present disclosure with reference to the accompanying drawings. All the above optional technical solutions may be combined arbitrarily to form the optional embodiments of the present disclosure, and are not described herein again.
The terms "step 1," "step 2," "step 3," and the like in the description and claims of this application and the above-described figures, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It should be understood that the data so used may be interchanged under appropriate circumstances such that the embodiments of the application described herein may be practiced in sequences other than those described herein.
The embodiment of the present disclosure provides a pedestrian re-identification data discrimination method based on an image quality discrimination model, fig. 1 is a flow chart of the pedestrian re-identification data discrimination method based on the image quality discrimination model provided by the embodiment of the present disclosure, and fig. 2 is a schematic diagram of the image quality discrimination model provided by the embodiment of the present disclosure, and in combination with the two diagrams, the method mainly includes the following steps:
step 1, determining an image quality evaluation standard;
and (3) formulating a pedestrian image quality evaluation standard according to the requirement of a pedestrian re-identification module, setting a threshold value, and dividing the image into a high-quality image and a low-quality image. Preferably, the pedestrian image quality evaluation criteria in step 1 include the image definition, whether noise exists, whether a pedestrian is complete and clear, whether the background is complex, the pedestrian image resolution, the shooting angle, the light source and the like.
Step 2, acquiring a pedestrian image, labeling the pedestrian and image quality according to the standard in the step 1 to form a sample set, and distinguishing a high-quality image and a low-quality image for subsequent training;
preferably, in step 2, in order to ensure the integrity of the sample set in the time dimension, when the images of the pedestrians are collected, a certain amount of images are respectively intercepted from the videos in four time periods of morning, noon, afternoon and evening.
Preferably, the acquiring of the pedestrian image further includes acquiring the pedestrian image output by the pedestrian detection module.
Step 3, preprocessing the collected pedestrian image, specifically comprising: the mean and variance of the image data in the sample set are calculated and the pixel values of the image are normalized to [0,1] and then scaled to the appropriate size, resulting in a high quality image data set and a low quality image data set.
Step 4, training an image quality discrimination model based on a reversible decoder by using the high-quality image data set obtained in the step 3; as shown in figure 1.
Reversible decoder training: using a reversible depth model i-RevNet as a decoder and the inverse of i-RevNet as an encoder, inputting the high-quality image data set in the step 3 into the encoder, connecting an output layer of the encoder with an input layer of the decoder, restoring the characteristics into an image by the decoder, calculating the performance of the decoder through a formula 1, and reducing L through multiple rounds of training softmax So that the decoder can learn the characteristics of a high-quality graph as much as possible to obtain the decoder for training the classifier;
equation 1:
wherein theta and omega are parameters required to be learned by the model, and high-quality graph characteristics
The characteristic x of the low-quality image is as a function of
y ∈ γ = {0,1, \ 8230;, c } represents a set of quantity categories in c,
is a plurality of
y, and H (a) is the feature of a extracted by the feature extractor.
Training a quality discrimination classifier: fixing the i-RevNet parameters of the decoder, selecting ResNet as an image quality classifier, inputting the high-quality image data set and the low-quality image data set in the step 3 into the classifier, connecting an output layer of the classifier with an input layer of the decoder, calculating the performance of the encoder and the classifier through a formula 2 for judging the training effect, obtaining the trained classifier after multiple rounds of training, and outputting the classifier as an image quality score.
Equation 2:
wherein
Is a parameter that needs to be trained at the current stage,
parameters fixed for the decoder;
the classifier is extracted as an image quality discrimination model based on a reversible decoder.
Because the decoder learns the characteristics of the high-quality image in advance, the judgment capability of the classifier on the quality is increased, the final score has higher credibility, and even can participate in the score calculation of pedestrian identification.
Step 5, accessing the image quality discrimination model based on the reversible decoder into the device;
and sending the image to be identified into an image quality judging model based on a reversible decoder to obtain a quality judging score. A pedestrian re-identification module calculates a final pedestrian identification result in combination with the quality score.
The present invention has been described in detail with reference to the accompanying drawings, and it is to be understood that the invention is not limited to the specific embodiments shown, but is intended to cover various insubstantial modifications of the invention based on the technical spirit and scope of the invention; or directly apply the conception and the technical scheme of the invention to other occasions without improvement and equivalent replacement, and the invention is within the protection scope of the invention.
Claims (4)
1. A pedestrian re-identification data discrimination method based on an image quality discrimination model is characterized by mainly comprising the following steps:
step 1, determining an image quality evaluation standard;
according to the requirements of a pedestrian re-identification module, formulating a pedestrian image quality evaluation standard, setting a threshold value, and dividing an image into a high-quality image and a low-quality image;
step 2, acquiring a pedestrian image, labeling the pedestrian and image quality according to the standard in the step 1 to form a sample set, and distinguishing a high-quality image and a low-quality image for subsequent training;
step 3, preprocessing the collected pedestrian image, specifically comprising: calculating the mean value and the variance of the image data in the sample set, normalizing the pixel value of the image to [0,1], and then scaling to a proper size to obtain a high-quality image data set and a low-quality image data set;
step 4, training an image quality discrimination model based on a reversible decoder by using the high-quality image data set obtained in the step 3;
reversible decoder training: using reversible depth modesThe type i-RevNet is used as a decoder, the inverse of the type i-RevNet is used as an encoder, the high-quality image data set in the step 3 is input into the encoder, an encoder output layer is connected with an decoder input layer, the decoder restores the characteristics into images, the performance of the decoder is calculated through a formula 1, and L is reduced through multiple rounds of training softmax So that the decoder can learn the characteristics of a high-quality graph as much as possible to obtain the decoder for training the classifier;
equation 1:
wherein theta and omega are parameters required to be learned by the model, and high-quality graph characteristics
The characteristic x of the low-quality image is as a function of
y ∈ γ = {0,1, \8230;, c } represents a set of quantity categories in c,
is a plurality of
y, and H (a) is the feature of a extracted by the feature extractor;
training a quality discrimination classifier: fixing the i-RevNet parameters of the decoder, selecting ResNet as an image quality classifier, inputting the high-quality image data set and the low-quality image data set in the step 3 into the classifier, connecting an output layer of the classifier with an input layer of the decoder, calculating the performance of the encoder and the classifier through a formula 2 for judging the training effect, obtaining the trained classifier after multiple rounds of training, and outputting the classifier as an image quality score;
equation 2:
wherein
Is a parameter that needs to be trained at the current stage,
parameters fixed for the decoder;
extracting a classifier as an image quality discrimination model based on a reversible decoder;
step 5, accessing the image quality discrimination model based on the reversible decoder into the device;
sending the image to be identified into an image quality discrimination model based on a reversible decoder to obtain a quality discrimination score; a pedestrian re-identification module calculates a final pedestrian identification result in combination with the quality score.
2. The method according to claim 1, wherein the pedestrian image quality assessment criteria in step 1 include one or more of image sharpness, noise, complete and clear pedestrian, complex background, pedestrian image resolution, shooting angle, and light source.
3. The method according to claim 1, wherein in step 2, in order to ensure the integrity of the sample set in the time dimension, a certain amount of images are respectively captured from videos of four time periods including morning, noon, afternoon and evening when the images of the pedestrians are collected.
4. The method according to any one of claims 1 to 3, wherein the step 2 of collecting the pedestrian image further comprises collecting the pedestrian image output by the pedestrian detection module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211059224.0A CN115423770A (en) | 2022-08-31 | 2022-08-31 | Pedestrian re-identification data discrimination method based on image quality discrimination model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211059224.0A CN115423770A (en) | 2022-08-31 | 2022-08-31 | Pedestrian re-identification data discrimination method based on image quality discrimination model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115423770A true CN115423770A (en) | 2022-12-02 |
Family
ID=84200930
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211059224.0A Withdrawn CN115423770A (en) | 2022-08-31 | 2022-08-31 | Pedestrian re-identification data discrimination method based on image quality discrimination model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115423770A (en) |
Citations (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110263697A (en) * | 2019-06-17 | 2019-09-20 | 哈尔滨工业大学（深圳） | Pedestrian based on unsupervised learning recognition methods, device and medium again |
| CN110414368A (en) * | 2019-07-04 | 2019-11-05 | 华中科技大学 | A kind of unsupervised pedestrian recognition methods again of knowledge based distillation |
| CN111612741A (en) * | 2020-04-22 | 2020-09-01 | 杭州电子科技大学 | An accurate no-reference image quality assessment method based on distortion identification |
| CN112200010A (en) * | 2020-09-15 | 2021-01-08 | 青岛邃智信息科技有限公司 | Face acquisition quality evaluation strategy in community monitoring scene |
| CN112365551A (en) * | 2020-10-15 | 2021-02-12 | 上海市精神卫生中心（上海市心理咨询培训中心） | Image quality processing system, method, device and medium |
| CN113313169A (en) * | 2021-05-28 | 2021-08-27 | 中国人民解放军战略支援部队航天工程大学 | Training material intelligent identification method, device and equipment based on deep learning |
| CN113408351A (en) * | 2021-05-18 | 2021-09-17 | 河南大学 | Pedestrian re-recognition method for generating confrontation network based on attitude guidance |
| CN113627383A (en) * | 2021-08-25 | 2021-11-09 | 中国矿业大学 | Pedestrian loitering re-identification method for panoramic intelligent security |
| WO2022001137A1 (en) * | 2020-06-29 | 2022-01-06 | 浪潮电子信息产业股份有限公司 | Pedestrian re-identification method, apparatus and device and medium |
| CN114495163A (en) * | 2022-01-13 | 2022-05-13 | 西安电子科技大学 | A Generative Learning Method for Person Re-ID Based on Category Activation Mapping |
| CN114494713A (en) * | 2020-11-13 | 2022-05-13 | 三星电子株式会社 | Method and apparatus for image recognition |
| CN114549365A (en) * | 2022-03-03 | 2022-05-27 | 重庆邮电大学 | Unmanned airborne high-fidelity SAR image reconstruction method for target detection |
-
2022
- 2022-08-31 CN CN202211059224.0A patent/CN115423770A/en not_active Withdrawn
Patent Citations (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110263697A (en) * | 2019-06-17 | 2019-09-20 | 哈尔滨工业大学（深圳） | Pedestrian based on unsupervised learning recognition methods, device and medium again |
| CN110414368A (en) * | 2019-07-04 | 2019-11-05 | 华中科技大学 | A kind of unsupervised pedestrian recognition methods again of knowledge based distillation |
| CN111612741A (en) * | 2020-04-22 | 2020-09-01 | 杭州电子科技大学 | An accurate no-reference image quality assessment method based on distortion identification |
| WO2022001137A1 (en) * | 2020-06-29 | 2022-01-06 | 浪潮电子信息产业股份有限公司 | Pedestrian re-identification method, apparatus and device and medium |
| CN112200010A (en) * | 2020-09-15 | 2021-01-08 | 青岛邃智信息科技有限公司 | Face acquisition quality evaluation strategy in community monitoring scene |
| CN112365551A (en) * | 2020-10-15 | 2021-02-12 | 上海市精神卫生中心（上海市心理咨询培训中心） | Image quality processing system, method, device and medium |
| US20220156888A1 (en) * | 2020-11-13 | 2022-05-19 | Samsung Electronics Co., Ltd. | Method and apparatus with image recognition |
| CN114494713A (en) * | 2020-11-13 | 2022-05-13 | 三星电子株式会社 | Method and apparatus for image recognition |
| CN113408351A (en) * | 2021-05-18 | 2021-09-17 | 河南大学 | Pedestrian re-recognition method for generating confrontation network based on attitude guidance |
| CN113313169A (en) * | 2021-05-28 | 2021-08-27 | 中国人民解放军战略支援部队航天工程大学 | Training material intelligent identification method, device and equipment based on deep learning |
| CN113627383A (en) * | 2021-08-25 | 2021-11-09 | 中国矿业大学 | Pedestrian loitering re-identification method for panoramic intelligent security |
| CN114495163A (en) * | 2022-01-13 | 2022-05-13 | 西安电子科技大学 | A Generative Learning Method for Person Re-ID Based on Category Activation Mapping |
| CN114549365A (en) * | 2022-03-03 | 2022-05-27 | 重庆邮电大学 | Unmanned airborne high-fidelity SAR image reconstruction method for target detection |
Non-Patent Citations (1)
| Title |
|---|
| 王亚;朱明;刘成林;: "基于CNN的监控视频中人脸图像质量评估", 计算机系统应用, no. 11, 14 November 2018 (2018-11-14) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108918532B (en) | System and method for detecting damage of expressway traffic sign | |
| CN110728694B (en) | Long-time visual target tracking method based on continuous learning | |
| CN112766056B (en) | Method and device for detecting lane lines in low-light environment based on deep neural network | |
| CN113139501B (en) | Pedestrian multi-attribute identification method combining local area detection and multi-level feature grabbing | |
| CN113569835A (en) | Water meter numerical value reading method based on target detection and segmentation identification | |
| CN115063679B (en) | Pavement quality assessment method based on deep learning | |
| Surekha et al. | AUTOMATIC LICENSE PLATE RECOGNITION USING IMAGE PROCESSING AND NEURAL NETWORK. | |
| CN102663362A (en) | Moving target detection method t based on gray features | |
| CN113280820A (en) | Orchard visual navigation path extraction method and system based on neural network | |
| CN111950409A (en) | A method and system for intelligent identification of road marking lines | |
| CN114973207A (en) | Road sign identification method based on target detection | |
| CN118230045A (en) | A garbage classification system and method based on visual recognition technology | |
| CN117690086A (en) | Flood prevention gate identification and control method and system based on 5G and AI technology | |
| CN111540203B (en) | Method for adjusting green light passing time based on fast-RCNN | |
| CN116665087A (en) | Method and device for identifying opening and closing states of disconnecting link of transformer substation | |
| CN119723643B (en) | A method and device for recognizing facial expression | |
| CN110674676A (en) | Road confidence estimation fuzzy frame method based on semantic segmentation | |
| CN115423770A (en) | Pedestrian re-identification data discrimination method based on image quality discrimination model | |
| CN113705348A (en) | Pedestrian re-identification method integrating multiple attention mechanisms | |
| CN117523521B (en) | A vehicle detection method based on Haar features and improved HOG features | |
| CN119478856A (en) | An artificial intelligence image processing method based on smart transportation | |
| CN113052011A (en) | Road target flow monitoring system based on computer vision | |
| CN112200008A (en) | Face attribute recognition method in community monitoring scene | |
| CN117036992A (en) | Method for obtaining water ecological health condition by image processing of unmanned aerial vehicle acquired image | |
| CN113689399B (en) | A remote sensing image processing method and system for power grid identification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20221202 |