Unmanned aerial vehicle video multi-target tracking method based on attention feature fusion

Abstract

The invention discloses an unmanned aerial vehicle video multi-target tracking method based on attention feature fusion, and aims to solve the problems of interference, shielding, variable viewpoint height and angle and the like in an unmanned aerial vehicle multi-target tracking video due to complex background factors. An unmanned aerial vehicle multi-target tracking algorithm network model is constructed and trained, a backbone network of the network selects a 34-layer residual network, a feature extraction network combined with a triple attention mechanism is designed in a feature extraction part, and a cascade feature fusion module is designed in an up-sampling part; the multi-target expression ability is stronger due to optimized features brought by the attention mechanism designed by the invention, the small targets in the unmanned aerial vehicle aerial video can be tracked more favorably due to the designed multi-scale information fusion channel, and the accuracy of predicting the multi-target trajectory in the unmanned aerial vehicle video by the association algorithm is further improved due to the optimized features.

Description

technical field [0001] The invention relates to a video multi-target tracking method, which integrates advanced technologies in many fields such as image processing, pattern recognition, artificial intelligence, automatic control, and computer, and particularly relates to a UAV video multi-target tracking algorithm based on attention feature fusion . Background technique [0002] Compared with manned aircraft, unmanned aerial vehicles (UAVs) are widely used in military and civilian fields due to their advantages such as small size, strong concealment, fast response, low requirements on the combat environment, and quick arrival at the scene. The wide application of UAVs requires target tracking technology, which can greatly enhance the autonomous flight and monitoring capabilities of UAVs, enabling UAVs to complete more types of tasks and adapt to more complex and changeable environments. Therefore, it is of great significance to study an effective and stable multi-target tr...

AI Technical Summary

Problems solved by technology

[0006] The existing UAV multi-target tracking algorithm has achieved good results for multiple human targets or vehicle targets under fixed shooting angles. Problems such as target loss caused by occlusion, shooting angle and height change

Aiming at problems such as complex background factor interference, occlusion, viewpoint height and angle change in UAV multi-target tracking video, the present invention proposes a UAV multi-target tracking algorithm based on attention feature fusion

Images