NT-VOT211: A Large-Scale Benchmark for Night-time Visual Object Tracking

Yu Liu, Arif Mahmood, Muhammad Haris Khan; Proceedings of the Asian Conference on Computer Vision (ACCV), 2024, pp. 194-212

Many current tracking benchmarks, such as OTB100, Nfs, UAV123, LaSOT, and GOT-10K, are approaching saturation i.e., they are approaching their maximum score capacity. As such, the leap in research progress towards realizing a robust and accurate tracking algorithm is mainly obstructed by the lack of a large-scale, well-annotated low-light condition dataset for rigorously benchmarking tracking algorithms. To this end, this paper presents NTB2024, a new benchmark dataset tailored for evaluating visual object tracking algorithms in the challenging low-light conditions. The dataset consists of 211 detailed videos, offering 211,000 annotated frames. It is among the largest tracking benchmark to-date that is specifically designed to address unique challenges such as adverse visibility, image blur, and low discrimination inherent to nighttime tracking scenarios. Through a comprehensive analysis of results obtained from 42 diverse tracking algorithms on NTB2024, we uncover the strengths and limitations of these algorithms, highlighting opportunities for enhancements in visual tracking, particularly in environments with suboptimal lighting. Besides, we public a leaderboard for revealing performance rankings, annotation tools, comprehensive meta-information and all the necessary code for reproducibility of results. We believe that our NTB2024 benchmark will be instrumental in facilitating visible scientific achievements and will also unlock avenues for new real-world tracking applications.