Visual Tracking

This repository records the proposed visual tracking methods based on PyTracking Library

Trackers

Here is a list of the improved trackers.

Models & Raw Results (Google Driver) Models & Raw Results (Baidu Driver: VisT)

• JDTrack: High-Performance Discriminative Tracking with Spatio-Temporal Template Fusion, ACM MM, 2025. [Paper]
• PrompT: Learning Color Prompt and Position Constraint for Visual Tracking, Engineering Applications of Artificial Intelligence, 2025. [Paper]
• CAT: Color Attention Tracking with Score Matching, International Journal of Machine Learning and Cybernetics, 2024. [Paper]
• FuDiMP: Attention Fusion and Target-Uncertain Detection for Discriminative Tracking, Knowledge-Based Systems, 2023. [Paper]
• EnDiMP: Enhancing Discriminative Appearance Model for Visual Tracking, Expert Systems With Applications, 2023. [Paper]
• RVT: Exploring Reliable Visual Tracking via Target Embedding Network, Knowledge-Based Systems, 2022. [Paper]
• OUPT: Learning Object-Uncertainty Policy for Visual Tracking, Information Sciences, 2022. [Paper]
• VSLT: Variable Scale Learning for Visual Object Tracking, Journal of Ambient Intelligence and Humanized Computing, 2021. [Paper]

Highlights

Moreover, we also integrate got10k and pysot_toolkit for evaluation.

The refine_modules of AlphaRefine tracker also is attached to adapt for bounding-box to mask output.

ECO, DiMP, TOMP, D3S, SiamFC, TranT, and TrDiMP trackers are seamlessly integrated into VisTrack project.

Findings

• The UAV123 results of pysot and pytracking are consistent, which are lower than the results of got10k toolkit. From the results reported in the original conference papers, the authors probably use the evaluation results given by got10k.

Contact

• Xuedong He (email:hexuedong@zjnu.edu.cn)
  Feel free to contact me if you have additional questions.