Target tracking method and system based on unmarked video training, terminal and medium

Abstract

The invention provides a target tracking method and system based on unmarked video training, and the method comprises the steps: carrying out the unsupervised optical flow prediction of an original video, extracting a candidate frame of each frame in the original video, and obtaining a candidate frame sequence; constructing a pseudo calibration frame sequence of a moving object in the original video based on the candidate frame sequence; constructing a training sample based on the pseudo calibration frame sequence, inputting the training sample into a naive twinning network to train the naive twinning network, and generating a preliminary tracking model; performing storage loop training on the preliminary tracking model to obtain a target tracking model; and tracking a target in a to-be-tracked video by using the target tracking model. Meanwhile, the invention provides a corresponding terminal and a medium. The cost of manual annotation of the video data is greatly reduced, and available video data is enriched and trained; and under the condition of no annotation, a target tracking model based on calibration frame regression is trained from an unannotated video.

Description

technical field [0001] The present invention relates to target tracking technology, in particular to a target tracking method, system, terminal and medium based on unlabeled video training. Background technique [0002] Object tracking is one of the most basic and important research directions in the field of computer vision. At present, target tracking technology has been widely used in intelligent warehouse management, live broadcast of sports events, unmanned aerial vehicles and other fields. Among them, the target tracking algorithm based on deep learning has attracted widespread attention from industry and academia due to its good generalization and high tracking accuracy. This type of deep learning-based target tracking algorithm is often trained based on supervised learning, which requires a large number of video datasets with accurate object location annotations as training samples. However, manual labeling of video information is time-consuming and laborious, and ...

AI Technical Summary

Problems solved by technology

However, there are some inherent defects in the UDT series of algorithms

First of all, when the UDT series uses cycle consistency for learning, the initial position of the tracking is random, so that UDT often starts from the background rather than the foreground object; and the lack of good initial frames also makes it difficult for the UDT series to learn a calibration frame regression model. thus limiting the upper performance limit of the UDT series

Second, when the UDT series algorithm is used for tracking learning, it can only learn circular tracking at short frame intervals based on the continuity of the video, so it is difficult to learn the large-scale movement and deformation of objects in long-term intervals.

Third, the UDT series did not focus on designing a deep network-based online tracking module

[0005] As mentioned above, the existing technology still has technical problems such as the inability to perform calibration frame regression, the inability to excavate the large deformation of moving objects, and the lack of training to update the tracking module online. At present, no description or report of similar technology to the present invention has been found, nor has it been collected domestically. similar information

Images