Unsupervised domain false label correction method and device in personnel re-identification
A re-identification and pedestrian re-identification technology, applied in the field of deep learning, can solve problems such as the recognition rate cannot meet the requirements, target occlusion, insufficient training data sets, etc., achieve fast and efficient recognition and detection, and improve feature expression and feature extraction capabilities Effect
Inactive Publication Date: 2021-07-30
TSINGHUA UNIV
View PDF6 Cites 1 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0004] This application provides an unsupervised field correcting pseudo-label method and device in person re-identification, to at least solve the current problems of camera shooting angle, picture imaging quality, environmental light changes, human body posture changes, target occlusion and insufficient training data sets. Influence, the problem that the recognition rate cannot meet the requirements
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0033] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some, not all, embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.
[0034] Based on the content in the background technology, the traditional pedestrian re-identification method is to use a supervised training method, that is, to use a manually collected and marked pedestrian training set for training, because the pictures of the training set are a limited number of collections in a limited scene, It does not fully conform to all scenes in real life. The trained pedestrian re-identification model has good detection and recognit...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention provides an unsupervised domain false label correction method and device in personnel re-identification, and realizes identification of a pedestrian picture without a label under the condition that only a scientific public label data set is used. The unsupervised learning algorithm facilitates the deployment of a pedestrian re-recognition system in a new data set, does not need a large amount of data labeling work in a new environment, and only needs to automatically detect pedestrians through a DMP algorithm and store pedestrian pictures for unsupervised training. According to the invention, the effect close to supervised learning training can be achieved, the workload of manual labeling is reduced, and rapid deployment and use of a pedestrian re-recognition system in a new environment are facilitated.
Description
technical field [0001] The present application belongs to the technical field of deep learning, and in particular, relates to an unsupervised domain correcting pseudo-label method and device in person re-identification. Background technique [0002] In the current state of the art, the purpose of person re-identification (re-ID) is to match an image of a person in one camera with images of that person in other cameras. Although supervised learning-based re-ID models have achieved a series of successes on several public re-identification datasets, when applied to new camera systems, they usually manifest as a significant performance drop in dataset domain shift, or by Annotate a large number of images on the image to maintain performance. New camera system. To reduce the reliance on labeled data, more and more person re-identification researchers have started to focus on the field of unsupervised learning. Unlike other unsupervised learning fields, where contrastive learni...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06N3/088G06V20/53G06V10/44G06N3/045G06F18/23
Inventor 丁贵广徐同坤
Owner TSINGHUA UNIV