CTIFTrack: Continuous Temporal Information Fusion for object track

Zhiguo Zhang, Zhiqing Guo, Liejun Wang, Yongming Li

Abstract: Highlights•A video-level linear temporal information fusion method is proposed for tracking task.•A continuous spatiotemporal mapping module is proposed for video-level features.•A multi-receptive field object template feature refinement module is proposed.•The proposed model has excellent tracking ability in the scene of object occlusion.•Our model has remarkable robustness and generalization ability.