CN112149538A - Pedestrian re-identification method based on multi-task learning - Google Patents
Pedestrian re-identification method based on multi-task learning Download PDFInfo
- Publication number
- CN112149538A CN112149538A CN202010960694.9A CN202010960694A CN112149538A CN 112149538 A CN112149538 A CN 112149538A CN 202010960694 A CN202010960694 A CN 202010960694A CN 112149538 A CN112149538 A CN 112149538A
- Authority
- CN
- China
- Prior art keywords
- pedestrian
- attribute
- identity
- loss
- network
- 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- Withdrawn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; CALCULATING OR 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/292—Multi-camera tracking
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a pedestrian re-identification method based on multi-task learning, which comprises the following steps: acquiring a pedestrian image of a cross-camera, and constructing a pedestrian re-identification training data set; constructing a multi-task learning network, wherein the network can combine an attribute task and an identity recognition task, and jointly optimize all parameters and loss, so that the aim of improving the accuracy rate of pedestrian re-recognition is fulfilled; the attribute task and the identification task respectively comprise verification and verification steps, the distance of the sample is optimized through classification and verification loss, the identification loss is used for constructing a large class space, and meanwhile, the loss is verified.
Description
Technical Field
The invention relates to a pedestrian re-identification method, in particular to a pedestrian re-identification method based on multi-task learning, and belongs to the technical field of deep learning.
Background
The pedestrian re-identification is to identify pedestrian images shot by a plurality of overlapped cameras, a target pedestrian shot by one camera is used as a retrieval object, the pedestrian is accurately identified from images shot by other cameras, and the pedestrian re-identification consists of two processes of extracting robust representation characteristics and adopting an effective measurement model to realize identification.
Prior to the advent of deep learning techniques, early pedestrian re-identification studies focused primarily on how to manually design better visual features and how to learn better similarity measures. With the development of deep learning in recent years, deep learning technology is widely applied to the task of pedestrian re-identification. Different from the traditional method, the deep learning method can automatically extract better pedestrian image characteristics, and simultaneously learn to obtain better similarity measurement.
The pedestrian re-identification is influenced by various condition changes, the identification precision is still low, and new technologies and algorithms with better robustness need to be provided. The performance of single and single-level image representation features is different in different scenes, and the image from different cameras cannot be well recognized by performing projection transformation learning on the image global or local structure information through a subspace technology with strong discrimination capability.
The specific reason is analyzed as follows:
1) the representation features of different levels and different parts represent different semantics, and the image is described by different views, different levels, different local visual color features, structural features or depth features and the like, and the different semantics are reflected, research results show that the recognition effect is not as good as that of multiple characteristics and multi-layer characteristic fusion effect by using a single type to express the characteristics and a single layer to express the characteristics, but the multiple characteristics are fused and utilized without distinction, the robustness is poor, a feature fusion utilization technology with strong environmental adaptability needs to be researched, under different scenes (namely different complex changes), the effect of more robust features on the scenes is amplified, and the effect of the features with poor robustness on the scenes is reduced.
2) The deep convolutional neural network has excellent performance in image classification, learns the depth characteristics of pedestrians by utilizing the deep convolutional neural network, has good discrimination capability and good robustness to noise, but the network training needs a large amount of marking data, the marking data is time-consuming and labor-consuming, and how to increase the training data of the deep neural network is needed to be researched to improve the robustness of the depth characteristics.
3) Intra-class and inter-class differences between samples: intra-class differences and inter-class confusion are common problems for pedestrian re-identification datasets. Intra-class differences refer to the diversity, such as pose, angle, appearance, etc., that a pedestrian is allowed to assume under different cameras, and inter-class confusion refers to the similar appearance of different pedestrians under cameras. The reason for these sample differences is the complexity across cameras, and the difficulty of learning and matching these features by the model.
Disclosure of Invention
The invention aims to provide a pedestrian re-identification method based on multi-task learning, which can be used for aiming at the problems that local information is not considered when only the overall outline of a pedestrian is identified in an identification task and attribute verification is not considered in an attribute task, and the identification accuracy is improved.
The purpose of the invention is realized as follows: a pedestrian re-identification method based on multitask learning comprises the following steps:
step 1) acquiring a cross-camera pedestrian image, and constructing a pedestrian re-identification training data set, wherein the data set comprises a preset number of pedestrian images;
step 2) taking ResNet-50 as a backbone network, and pre-training a pre-training model based on ImageNet on pedestrian re-recognition data to obtain a pedestrian re-recognition pre-training network model; respectively carrying out fine adjustment on the pedestrian re-identification pre-training network by adopting an identity tag and an attribute tag to generate an identity identification network and an attribute identification network;
step 3) dividing the identity recognition network into an identity classification task and an identity verification task, dividing the attribute recognition network into an attribute classification task and an attribute verification task, performing identity prediction by using a loss function in the identity classification task, and calculating classification loss for a feature vector output by a sample picture;
step 4) in the authentication task, giving two pictures, inputting the two pictures into a pedestrian re-recognition pre-training network model, acquiring global feature vectors of the two pictures, and calculating a loss function of the picture pair; selecting N image pairs, and calculating the total verification loss;
step 5), in the attribute classification task, performing attribute prediction by using M loss functions, and calculating the loss of M attributes of M networks;
step 6), the attribute verification task is to use a comparison loss function to calculate the total loss of the attribute verification task in order to judge whether the extracted attribute features accord with the pedestrian features;
step 7) the total loss generated by the four tasks is adjusted by balancing parameters alpha and beta, wherein the parameter alpha balances the contribution between the identity and the attribute, and the beta balances the contribution between the classification and the verification.
As a further improvement of the present invention, step 1) specifically comprises:
1.1) detecting a pedestrian image foreground by adopting a Gaussian mixture model;
1.2) according to the detection result of the step 1.1, if a video frame with a moving foreground exists, detecting the pedestrian by adopting a pre-trained pedestrian detector, accurately positioning the position of the pedestrian, and intercepting a corresponding area image from the video frame as a pedestrian image;
1.3) according to the detection result of the step 1.1, if the Gaussian mixture model does not detect the motion foreground, the pedestrian detector is not executed;
1.4) marking the same pedestrian extracted from different cameras as the same class by adopting a manual marking mode, and giving a serial number, wherein the image of each class of pedestrian is not less than the preset number of samples; different numbers are used for representing different pedestrian images; manually labeling the attribute label of each pedestrian; the sample collection process described above is iterated, and data collection may be stopped when the training data set size contains a preset number of pedestrian images.
As a further improvement of the present invention, step 2) specifically includes:
2.1) obtaining a deep convolution network model trained in advance on the ImageNet data set, and training the deep convolution network model on the pedestrian re-identification data;
2.2) when the deep convolutional neural network model is pre-trained on pedestrian re-identification data, the deep convolutional network model is respectively subjected to fine tuning by utilizing sample identity information and attribute information, an average pooling layer is added to a ResNet50 network model pre-trained on an ImageNet data set, and a dropout layer is inserted behind the average pooling layer.
As a further improvement of the present invention, step 3) specifically includes:
3.1) acquiring an image group from the data set, wherein the image group comprises an identity tag and an attribute tag;
3.2) using a loss function to predict the identity, and acquiring the prediction probability of the global image of the pedestrian according to the identity recognition classification network;
3.3) defining a loss function in the identity classification network according to the prediction probability.
As a further improvement of the present invention, step 4) specifically includes:
4.1) determining two images I, J, extracting the corresponding advanced features f of the images through an identity recognition networka,fb；
4.2) according to the characteristics extracted in the step 4.1, carrying out similarity estimation by using a contrast loss function;
4.3) calculating the verification loss of N image pairs in each batch according to the loss function of a group of images.
As a further improvement of the present invention, step 5) specifically includes:
5.1) given a sample image I, assume that there are m classes of attributes;
5.2) performing attribute prediction by using a loss function, and acquiring the prediction probability of the pedestrian image attribute according to an attribute classification network;
5.3) defining a loss function in the attribute classification network according to the prediction probability.
As a further improvement of the present invention, step 7) specifically includes:
7.1 combining the four losses generated in the step 3), the step 4), the step 5) and the step 6) to generate a final multitasking loss;
7.2 the loss between identity and attribute is balanced using parameter α;
7.3 the loss between classification and validation is balanced using the parameter β.
Compared with the prior art, the invention adopting the technical scheme has the following technical effects: the existing deep learning end-to-end learning framework is improved, an attribute task and an identity recognition task are combined, a multi-task learning network is provided, pedestrian re-identification is a task of finding an inquired person from non-overlapping cameras, and the target is to find the inquired person; the purpose of attribute identification is to predict the presence of a set of attributes from an image; attributes describe a person's details including gender, accessories, clothing color, etc.; the purpose of the recognition task is to determine whether the person is the person we want to query; by utilizing the attribute labels, the model can learn to classify the pedestrians by clearly paying attention to some local semantic descriptions, so that the training of the model is greatly simplified, and the re-identification performance of large-scale people is improved by utilizing complementary clues in the attribute labels; combining the pedestrian attribute task with the pedestrian identification task, and providing supplementary information from different angles by integrating multi-context information; the attribute is focused on local information of one person, the identity is focused on the overall outline and the appearance, the identification loss is used for constructing a large class space, and the verification loss is used for optimizing the space, so that the distance between similar pictures is smaller, and the distance between different pictures is larger; the two classification and verification methods are combined in anticipation, two tasks of identity and attribute are fully utilized, and the verification technology is embedded into the attribute, so that the defect of an attribute identification model is made up to a certain extent.
Drawings
FIG. 1 is a diagram of an overview of a multitask learning method according to an embodiment of the present invention.
Fig. 2 is an architecture of a multitask learning network based on an embodiment of the present invention.
Detailed Description
The technical solution of the present invention is further described in detail below with reference to the accompanying drawings.
The pedestrian re-identification method based on the multi-task learning shown in fig. 1 comprises the following steps.
Step 1) acquiring a cross-camera pedestrian image, and constructing a pedestrian re-identification training data set, wherein the data set comprises a preset number of pedestrian images;
in the embodiment, a Gaussian mixture model is adopted to detect the foreground of the pedestrian image; according to the detection result of the Gaussian mixture model, if a video frame with a moving foreground exists, detecting the pedestrian by adopting a pre-trained pedestrian detector, accurately positioning the position of the pedestrian, and intercepting a corresponding region image from the video frame as a pedestrian image; if the Gaussian mixture model does not detect the motion foreground, the pedestrian detector is not executed; marking the same pedestrian extracted from different cameras as the same class by adopting a manual marking mode, and giving a serial number, wherein the number of each class of pedestrian image is not less than the preset number of samples; different numbers are used for representing different pedestrian images; manually labeling the attribute label of each pedestrian; the sample collection process described above is iterated, and data collection may be stopped when the training data set size contains a preset number of pedestrian images.
Step 2) taking ResNet-50 as a backbone network, and pre-training a pre-training model based on ImageNet on pedestrian re-recognition data to obtain a pedestrian re-recognition pre-training network model; respectively carrying out fine adjustment on the pedestrian re-identification pre-training network by adopting an identity tag and an attribute tag to generate an identity identification network and an attribute identification network;
in this embodiment, the networks are all convolutional neural networks, and each convolutional unit comprises a plurality of convolutional units and a pooling layer, wherein each convolutional unit comprises a batch normalization layer, a convolutional layer and a nonlinear activation layer; in recent years, a convolution network in deep learning shows a good effect in extracting high-level features of an image, but information extracted by a convolution kernel lacks enough target prior information; therefore, the attribute task is combined with the identification task, and the attribute task is used for supplementing the identity identification so as to improve the accuracy of pedestrian identification;
acquiring a deep convolution network model trained on an ImageNet data set in advance, and training the deep convolution network model on pedestrian re-identification data; when the deep convolutional neural network model is pre-trained on pedestrian re-identification data, the deep convolutional network model is respectively subjected to fine tuning by utilizing sample identity information and attribute information, an average pooling layer is added to a pre-trained ResNet50 network model on an ImageNet data set, and a dropout layer is inserted behind the average pooling layer; the two sub-networks of the identity network and the attribute network have the same structure and share parameters.
Step 3) dividing the identity recognition network into an identity classification task and an identity verification task, using a loss function to predict the identity in the identity classification task, and calculating classification loss by using a feature vector output by a sample picture;
in this embodiment, the multitask learning network can input a single picture for learning and output the global features of the picture; the global feature extracts identity features through an identity network layer FC _ IC layer, identity prediction is carried out by using a loss function, and similarity among samples is well expressed. I ═ xn,kn,anIs a group of pictures in which the identity label is knAnd attribute label anEach image xnA subset containing identity labels k e {1 … … k } and attribute labels
If the training picture is x, then output z ═ z1,z2……zk]The probability of each identity label k e {1,2 … … k } can be predicted as P, and the loss function is defined as
Wherein
Is the prediction probability, PiIs the target probability; except for Pi＝kAll p except 1i0, f is a 2048 dimensional feature.
Step 4) in the authentication task, giving two pictures, inputting the two pictures into a pedestrian re-recognition pre-training network model, acquiring global feature vectors of the two pictures, and calculating a loss function of the picture pair; selecting N image pairs, and calculating the total verification loss;
in the embodiment mode, the parameters of each layer of the objective function relative to the network are solved by utilizing network forward propagation, and the parameters of each layer are updated and learned by utilizing random gradient descent;
in this example, two images I, J are given, and the high-level feature f corresponding to the image is extracted through the identification network FC _ I layera,fbThe identity label corresponding to each picture is y. according to the extracted high-level features fa,fbPerforming similarity estimation by using a contrast loss function; the image pair loss function is defined as:
measuring distance by Euclidean distance, using | · |. non-woven phosphor1Represents; i | · | purple wind1A first norm representing a vector; | represents solving for absolute value, λ is a weighting parameter, controls the strength of the regularizer, θ>0 is a margin threshold parameter; in the present invention, we set the parameters λ 0.01 and θ 1024; if the two pictures are similar, y is 0; otherwise, y is 1; the first penalty item penalizes similar images mapped to different codes; the second penalty item punishs that when the Euclidean distance is lower than a threshold value theta, different images are mapped to similar image codes; the last term is used to match the output to the required discrete value (+1/-1), thus ensuring convergence of the network; assuming random selection of N pairs from a small batch, our final goal is to minimize the total loss function, which can be defined as Lm(fa,fbY), which is the sum of the loss functions from 1 to N.
Step 5), in the attribute classification task, performing attribute prediction by using M loss functions, and calculating the loss of M attributes of M networks;
in this example manner, similarly to step 3), I ═ { x) is givenn,kn,an}, and attribute tags
We use M loss functions for attribute prediction, and attribute i is assigned to attribute classThe probability of class j ∈ {1,2 … … m } is P, then
The cross-entropy cost function of an attribute classification can be defined as:
that is, the total attribute classification loss is the sum of each attribute classification loss. Suppose yi,mIs a true tag for attribute i; when j is equal to yi,mWhen is, Pi,j＝1。
Step 6), the attribute verification task is to use a comparison loss function to calculate the total loss of the attribute verification task in order to judge whether the extracted attribute features accord with the pedestrian features;
in this example approach, the attribute verification task may be implemented by combining multiple different contrast penalties. Similar to step 4), given I ═ { x ═ xn,kn,an}, and attribute tags
Assuming that the attribute i has j e {1,2 … m } sorts, we can define the attribute verification penalty as:
step 7) the total loss generated by the four tasks is adjusted by balancing parameters alpha and beta, wherein the parameter alpha balances the contribution between the identity and the attribute, and the beta balances the contribution between the classification and the verification;
in this example, the final multitasking loss is generated by combining the four losses generated in the above steps 3), 4), 5), and 6); the multi-task learning network can simultaneously predict identity and attribute after being trained; the final loss function L is the sum of the four loss function values with a certain weight, the parameter α balances the loss between identity and attribute, and β balances the loss between classification and verification.
The above description is only an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can understand that the modifications or substitutions within the technical scope of the present invention are included in the scope of the present invention, and therefore, the scope of the present invention should be subject to the protection scope of the claims.
Claims (7)
1. A pedestrian re-identification method based on multi-task learning is characterized in that: the method comprises the following steps:
step 1) acquiring a cross-camera pedestrian image, and constructing a pedestrian re-identification training data set, wherein the data set comprises a preset number of pedestrian images;
step 2) taking ResNet-50 as a backbone network, and pre-training a pre-training model based on ImageNet on pedestrian re-recognition data to obtain a pedestrian re-recognition pre-training network model; respectively carrying out fine adjustment on the pedestrian re-identification pre-training network by adopting an identity tag and an attribute tag to generate an identity identification network and an attribute identification network;
step 3) dividing the identity recognition network into an identity classification task and an identity verification task, dividing the attribute recognition network into an attribute classification task and an attribute verification task, performing identity prediction by using a loss function in the identity classification task, and calculating classification loss for a feature vector output by a sample picture;
step 4) in the authentication task, giving two pictures, inputting the two pictures into a pedestrian re-recognition pre-training network model, acquiring global feature vectors of the two pictures, and calculating a loss function of the picture pair; selecting N image pairs, and calculating the total verification loss;
step 5), in the attribute classification task, performing attribute prediction by using M loss functions, and calculating the loss of M attributes of M networks;
step 6), the attribute verification task is to use a comparison loss function to calculate the total loss of the attribute verification task in order to judge whether the extracted attribute features accord with the pedestrian features;
step 7) the total loss generated by the four tasks is adjusted by balancing parameters alpha and beta, wherein the parameter alpha balances the contribution between the identity and the attribute, and the beta balances the contribution between the classification and the verification.
2. The pedestrian re-identification method based on multitask learning according to claim 1, characterized by comprising the following steps: the step 1) specifically comprises the following steps:
1.1) detecting a pedestrian image foreground by adopting a Gaussian mixture model;
1.2) according to the detection result of the step 1.1, if a video frame with a moving foreground exists, detecting the pedestrian by adopting a pre-trained pedestrian detector, accurately positioning the position of the pedestrian, and intercepting a corresponding area image from the video frame as a pedestrian image;
1.3) according to the detection result of the step 1.1, if the Gaussian mixture model does not detect the motion foreground, the pedestrian detector is not executed;
1.4) marking the same pedestrian extracted from different cameras as the same class by adopting a manual marking mode, and giving a serial number, wherein the image of each class of pedestrian is not less than the preset number of samples; different numbers are used for representing different pedestrian images; manually labeling the attribute label of each pedestrian; the sample collection process described above is iterated, and data collection may be stopped when the training data set size contains a preset number of pedestrian images.
3. The pedestrian re-identification method based on multitask learning according to claim 1, characterized by comprising the following steps: the step 2) specifically comprises the following steps:
2.1) obtaining a deep convolution network model trained in advance on the ImageNet data set, and training the deep convolution network model on the pedestrian re-identification data;
2.2) when the deep convolutional neural network model is pre-trained on pedestrian re-identification data, the deep convolutional network model is respectively subjected to fine tuning by utilizing sample identity information and attribute information, an average pooling layer is added to a ResNet50 network model pre-trained on an ImageNet data set, and a dropout layer is inserted behind the average pooling layer.
4. The pedestrian re-identification method based on multitask learning according to claim 1, characterized by comprising the following steps: the step 3) specifically comprises the following steps:
3.1) acquiring an image group from the data set, wherein the image group comprises an identity tag and an attribute tag;
3.2) using a loss function to predict the identity, and acquiring the prediction probability of the global image of the pedestrian according to the identity recognition classification network;
3.3) defining a loss function in the identity classification network according to the prediction probability.
5. The pedestrian re-identification method based on multitask learning according to claim 1, characterized by comprising the following steps: the step 4) specifically comprises the following steps:
4.1) determining two images I, J, extracting the corresponding advanced features f of the images through an identity recognition networka,fb；
4.2) according to the characteristics extracted in the step 4.1, carrying out similarity estimation by using a contrast loss function;
4.3) calculating the verification loss of N image pairs in each batch according to the loss function of a group of images.
6. The pedestrian re-identification method based on multitask learning according to claim 1, characterized by comprising the following steps: the step 5) specifically comprises the following steps:
5.1) given a sample image I, assume that there are m classes of attributes;
5.2) performing attribute prediction by using a loss function, and acquiring the prediction probability of the pedestrian image attribute according to an attribute classification network;
5.3) defining a loss function in the attribute classification network according to the prediction probability.
7. The pedestrian re-identification method based on multitask learning according to claim 1, characterized by comprising the following steps: step 7) specifically comprises:
7.1 combining the four losses generated in the step 3), the step 4), the step 5) and the step 6) to generate a final multitasking loss;
7.2 the loss between identity and attribute is balanced using parameter α;
7.3 the loss between classification and validation is balanced using the parameter β.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010960694.9A CN112149538A (en) | 2020-09-14 | 2020-09-14 | Pedestrian re-identification method based on multi-task learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010960694.9A CN112149538A (en) | 2020-09-14 | 2020-09-14 | Pedestrian re-identification method based on multi-task learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112149538A true CN112149538A (en) | 2020-12-29 |
Family
ID=73892207
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010960694.9A Withdrawn CN112149538A (en) | 2020-09-14 | 2020-09-14 | Pedestrian re-identification method based on multi-task learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112149538A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112613474A (en) * | 2020-12-30 | 2021-04-06 | 珠海大横琴科技发展有限公司 | Pedestrian re-identification method and device |
| CN112784772A (en) * | 2021-01-27 | 2021-05-11 | 浙江大学 | In-camera supervised cross-camera pedestrian re-identification method based on contrast learning |
| CN112801051A (en) * | 2021-03-29 | 2021-05-14 | 哈尔滨理工大学 | Method for re-identifying blocked pedestrians based on multitask learning |
| CN113033428A (en) * | 2021-03-30 | 2021-06-25 | 电子科技大学 | Pedestrian attribute identification method based on instance segmentation |
| CN113128441A (en) * | 2021-04-28 | 2021-07-16 | 安徽大学 | System and method for identifying vehicle weight by embedding structure of attribute and state guidance |
| CN113807200A (en) * | 2021-08-26 | 2021-12-17 | 青岛文达通科技股份有限公司 | Multi-person identification method and system based on dynamic fitting multi-task reasoning network |
| CN115909464A (en) * | 2022-12-26 | 2023-04-04 | 淮阴工学院 | Self-adaptive weak supervision label marking method for pedestrian re-identification |
-
2020
- 2020-09-14 CN CN202010960694.9A patent/CN112149538A/en not_active Withdrawn
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112613474A (en) * | 2020-12-30 | 2021-04-06 | 珠海大横琴科技发展有限公司 | Pedestrian re-identification method and device |
| CN112613474B (en) * | 2020-12-30 | 2022-01-18 | 珠海大横琴科技发展有限公司 | Pedestrian re-identification method and device |
| CN112784772A (en) * | 2021-01-27 | 2021-05-11 | 浙江大学 | In-camera supervised cross-camera pedestrian re-identification method based on contrast learning |
| CN112784772B (en) * | 2021-01-27 | 2022-05-27 | 浙江大学 | In-camera supervised cross-camera pedestrian re-identification method based on contrast learning |
| CN112801051A (en) * | 2021-03-29 | 2021-05-14 | 哈尔滨理工大学 | Method for re-identifying blocked pedestrians based on multitask learning |
| CN113033428A (en) * | 2021-03-30 | 2021-06-25 | 电子科技大学 | Pedestrian attribute identification method based on instance segmentation |
| CN113128441A (en) * | 2021-04-28 | 2021-07-16 | 安徽大学 | System and method for identifying vehicle weight by embedding structure of attribute and state guidance |
| CN113128441B (en) * | 2021-04-28 | 2022-10-14 | 安徽大学 | System and method for identifying vehicle weight by embedding structure of attribute and state guidance |
| CN113807200A (en) * | 2021-08-26 | 2021-12-17 | 青岛文达通科技股份有限公司 | Multi-person identification method and system based on dynamic fitting multi-task reasoning network |
| CN113807200B (en) * | 2021-08-26 | 2024-04-19 | 青岛文达通科技股份有限公司 | Multi-row person identification method and system based on dynamic fitting multi-task reasoning network |
| CN115909464A (en) * | 2022-12-26 | 2023-04-04 | 淮阴工学院 | Self-adaptive weak supervision label marking method for pedestrian re-identification |
| CN115909464B (en) * | 2022-12-26 | 2024-03-26 | 淮阴工学院 | Self-adaptive weak supervision tag marking method for pedestrian re-identification |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112149538A (en) | Pedestrian re-identification method based on multi-task learning | |
| Li et al. | R $^ 3$-net: A deep network for multi-oriented vehicle detection in aerial images and videos | |
| CN111709311B (en) | Pedestrian re-identification method based on multi-scale convolution feature fusion | |
| Qu et al. | RGBD salient object detection via deep fusion | |
| Liu et al. | Fusion tracking in color and infrared images using joint sparse representation | |
| Yang et al. | Multi-object tracking with discriminant correlation filter based deep learning tracker | |
| CN110807434B (en) | Pedestrian re-recognition system and method based on human body analysis coarse-fine granularity combination | |
| CN111126360A (en) | Cross-domain pedestrian re-identification method based on unsupervised combined multi-loss model | |
| CN110580460A (en) | Pedestrian re-identification method based on combined identification and verification of pedestrian identity and attribute characteristics | |
| CN110633632A (en) | Weak supervision combined target detection and semantic segmentation method based on loop guidance | |
| Kaâniche et al. | Recognizing gestures by learning local motion signatures of HOG descriptors | |
| Ma et al. | Orientation driven bag of appearances for person re-identification | |
| CN110163117B (en) | Pedestrian re-identification method based on self-excitation discriminant feature learning | |
| Kobyshev et al. | Matching features correctly through semantic understanding | |
| Sahbi et al. | Frugal Learning for Interactive Satellite Image Change Detection | |
| CN110008899B (en) | Method for extracting and classifying candidate targets of visible light remote sensing image | |
| Jia et al. | Fabric defect detection based on transfer learning and improved Faster R-CNN | |
| Galiyawala et al. | Person retrieval in surveillance videos using deep soft biometrics | |
| Fritz et al. | Building detection from mobile imagery using informative SIFT descriptors | |
| Zhao et al. | Learning saliency features for face detection and recognition using multi-task network | |
| Fritz et al. | Object recognition using local information content | |
| Shf et al. | Review on deep based object detection | |
| Mokalla et al. | On designing MWIR and visible band based deepface detection models | |
| Jourdheuil et al. | Heterogeneous adaboost with real-time constraints-application to the detection of pedestrians by stereovision | |
| Li et al. | Smoking behavior recognition based on a two-level attention fine-grained model and EfficientDet network |
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: 20201229 |