Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Target tracking method based on task distinguishing detection and re-identification joint network

A re-identification and task technology, applied in the field of computer vision, can solve problems such as not being able to adapt to different tasks, and achieve the effect of improving accuracy

Pending Publication Date: 2022-04-12
BEIHANG UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Therefore, it is not enough to adopt the same network structure for these two tasks, and only use the constraints of different loss functions in the task network branch structure to train the exact same features extracted, which is not enough to adapt to different tasks.

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Target tracking method based on task distinguishing detection and re-identification joint network
  • Target tracking method based on task distinguishing detection and re-identification joint network
  • Target tracking method based on task distinguishing detection and re-identification joint network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0028] As mentioned above, the present invention proposes a target tracking method based on task differentiation, detection and re-identification joint network. The specific implementation of the present invention will be described below with reference to the accompanying drawings.

[0029] (1) Overall process

[0030] The invention proposes a joint network of task differentiation, detection and recognition to realize video multi-target tracking. The overall framework diagram of the joint network of task differentiation, detection and re-identification is shown in the attached figure 1 As shown, it mainly includes three parts: (1) backbone network; (2) multi-feature fusion network; (3) multi-task branch. These three parts are also three steps of the method proposed by the present invention.

[0031] For the input current frame, first use the DLA backbone network (as attached figure 2 (shown) extracts the shared features used in the target detection task and target re-ident...

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

No PUM Login to View More

Abstract

The invention provides a target tracking method based on a task distinguishing detection and re-identification combined network. According to the method, a task-differentiated multi-feature fusion target detection and re-identification combined network is constructed based on FairMOT, a target detection task and a target re-identification feature extraction task are integrated in the same combined network, and shared features are extracted by using a backbone network, so that differentiated multi-feature fusion is performed according to task features; according to the method, the emphasis of two tasks on feature requirements is fully considered while target detection and target re-identification feature extraction tasks are balanced, the accuracy of target detection and re-identification feature extraction is improved, and accurate multi-target tracking is further realized. Wherein a multi-task layered feature fusion structure or a multi-task independent feature fusion structure is adopted in the multi-feature fusion network, so that two different tasks can fuse information of different scales, task-oriented feature separation is realized earlier, and fusion features which are more beneficial to different sub-task branches are obtained.

Description

technical field [0001] The invention relates to a target tracking method based on task distinction, detection and re-identification joint network, which realizes tracking of multiple targets in video, and belongs to the technical field of computer vision. Background technique [0002] Video multi-target tracking is one of the important problems in the field of computer vision. Its main purpose is to locate the position of each target in the form of a target frame in each frame of the video, and then form the target trajectory. At the same time, the targets belonging to the same target boxes need to be given the same identity tag. Therefore, the essence of video multi-target tracking tasks is target location and target association. The former is mainly realized by target detection technology, while the latter is mainly realized by three key steps: target re-identification feature extraction, target similarity calculation and target matching. Target detection and re-identific...

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
no application Login to View More
IPC IPC(8): G06V20/40G06K9/62G06N3/04G06N3/08G06V10/80G06V10/774G06V10/764G06V10/82
Inventor 郑锦许银翠王念
Owner BEIHANG UNIV
Features
  • Generate Ideas
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More