The paper " MSITrack: A Challenging Benchmark for Multispectral Single Object Tracking " has been accepted by ACMMM 2025 Datasets Track.
MSITrack, a large-scale and challenging benchmark for Multispectral Single Object Tracking, has been released.
MSITrack is currently the largest and most diverse multispectral single object tracking dataset, comprising 300 videos and over 129k frames of multispectral imagery. Each frame contains 8 bands covering from visible light to near-infrared bands, with a spatial resolution of up to 1200×900 pixels. MSITrack includes a wide range of tracking categories. In addition to commonly seen classes such as pedestrian, car and bicycle, the dataset features less frequently encountered objects such as meerkat, lion, gorilla, flamingo, elk, monkey and giraffe, totaling 55 distinct categories. Many of these classes are introduced to the multispectral tracking domain for the first time. Furthermore, the dataset incorporates 11 key challenge attributes, making it highly representative of real world challenges encountered in practical applications.
Download link: https://pan.baidu.com/s/1meel7dxCliXb6YnFp8QyXw?pwd=msit
If you use this benchmark in your research, please cite this project.
This repository contains two components with different licenses:
Our code is released under the 
The MSITrack dataset is licensed under 