Online multi-target tracking method, system and application

Abstract

The invention belongs to the technical field of computer vision and deep learning, and discloses an online multi-target tracking method, system and application, and the method comprises the steps: inputting a current frame image of a video into a convolutional neural network; after convolution in the convolutional neural network, extracting features of different targets on different channel feature maps; fusing the extracted features into a feature matrix; inputting a next frame of image of the video, and repeating the above steps to obtain a feature matrix of the frame; performing data association operation on the feature matrix of the current frame and the previously obtained feature matrix of the previous n frames (n is greater than or equal to 1 and less than or equal to 30); and achieving correct tracking between targets by utilizing an improved Hungary algorithm according to a result after the data association operation, and achieving a multi-target tracking method. According tothe experimental result, the online multi-target tracking method effectively improves the tracking accuracy and has good robustness in a complex scene.

Description

technical field [0001] The invention belongs to the technical field of computer vision and deep learning, and in particular relates to an online multi-target tracking method, system and application. Background technique [0002] With the continuous development of artificial intelligence technology, more and more science and technology have been greatly updated, such as computer vision. Now the mainstream method is to use convolutional neural network to extract image features and use them to realize the next step such as tasks such as classification and detection. The same is true for the multi-target tracking to be done, making full use of the characteristics of the convolutional neural network to realize the tracking and calibration of the detection target. Multi-object tracking technology has notable applications in fields such as drones, autonomous driving, motion analysis, and education. Online multi-target tracking technology has a history of decades of development. I...

AI Technical Summary

Problems solved by technology

This method combines the advantages of deep learning and traditional algorithms to realize the tracking algorithm. Although the two main methods of prediction and feature extraction are cleverly combined to solve the problem of multi-target tracking, the tracking speed is slow due to the use of Kalman filtering. Shortcomings

[0006] Through the above analysis, the existing problems and defects of the existing technology are: the extraction of target features by traditional methods mainly relies on manual calibration, the extracted features are poor in effect, and the tracking accuracy is low; the deep learning method is in its infancy, and there will be real-time Weak tracking ability, tracking target drift and other issues

Images