Abstract
Model update is an important module in target trackers. It plays an important role in adaptive tracking. Many researches have proven that different model update strategies should be adopted, when tracking in different scenes, especially in occlusion and deformation. Though many strategies have been proposed in recent years, few of them make high improvement and good combination on trackers. In this paper, we first proved there is a close relationship between the tracking scenes and the response maps. Then, we proposed an adaptive model update strategy for calculating model update rate based on the response map. Many experiments have been done to compare the proposed model update strategy with some state-of-the-art strategies, and the results have shown that the proposed model update strategy outperforms the best model update strategy by 7% on the test of Kernel Correlation Filter tracker. Furthermore, the proposed model update strategy was evaluated on some state-of-the-art correlation filter trackers. Results have shown the proposed strategy was well integrated into many trackers, and improved the tracking accuracy effectively.
Highlights
Visual tracking is a fundamental task for computer vision with wide applications
The upper right corner shows the area under curve (AUC) of the trackers which is the area enclosed by the curves and the xy axis
Which show that the proposed model update strategy has improved the tracking accuracy of staple, DSST and siamfc in almost all of the tracking problems
Summary
Visual tracking is a fundamental task for computer vision with wide applications. Domains such as video surveillance, human-computer interaction and smart car. Changes during tracking, so that to continuously update the model to adapt to the target. Model update is a common strategy for trackers to achieve adaptive tracking. The original model is updated linearly with a new model, which is generated based on the tracking result in the current frame This update strategy can achieve good results under normal scenes. When the update rate is set too high, error information can be involved in the target model. High update rate should be adopted when the tracking is in normal, or the scenes which target model need to keep up with the changes of the target quickly. An adaptive model update strategy was proposed in this manuscript to select appropriate model update rate for different tracking scenes.
Sign-in/Register to view highlights & summary 
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call