Visible light infrared target tracking training method and device based on non-paired data

Abstract

The invention discloses a visible light infrared target tracking training method and device based on non-paired data, and the method comprises the steps: obtaining a visible light image and a thermal infrared image which are not paired, and generating candidate samples; the visible light infrared tracker is trained through the candidate samples, the visible light infrared tracker comprises a modal specific module, a modal sharing module, a modal self-adaptive attention module and a modal adaptation module which are connected in sequence, and the modal specific module comprises a first modal specific network and a second modal specific network; the visible light image serves as the input of the first modal specific network and the modal sharing module, the thermal infrared image serves as the input of the second modal specific network and the modal sharing module, and the output of the first modal specific network and the output of the second modal specific network are fused with the output of the modal sharing module and then serve as the input of the modal adaptive attention module. The method gets rid of dependence on large-scale registration data, and improves the target tracking performance.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a visible light infrared target tracking training method and device based on unmatched data. Background technique [0002] Object tracking has always been an important topic in the field of computer vision. In recent years, object tracking technology has made great breakthroughs and is widely used in the fields of intelligent transportation, unmanned driving and robots. The task of target tracking is to predict the size and position of the target in subsequent frames given the size and position of the target in the initial frame of a video sequence. [0003] Most target tracking algorithms are based on a single visible light mode condition, and have superior tracking performance under this mode condition, but in complex environments or extreme conditions, such as haze weather, low light, tracking robustness There is still room for improvement. In recent years, more and ...

AI Technical Summary

Problems solved by technology

[0005] The technical problem to be solved by the present invention is how to solve the problem of dependence of visible light infrared target tracking performance training on large-scale registration data

Images