Researchers have introduced DRGBT-1K, a new large-scale benchmark designed to evaluate the robustness of dynamic RGBT tracking systems. This benchmark comprises over 1,000 real-world sequences and nearly 800,000 RGBT frame pairs, captured using unmanned aerial vehicles and handheld devices. It features comprehensive annotations including bounding boxes, target categories, modality labels, and platform labels, covering diverse scenes and challenges. The dataset also includes an unaligned version, UGVT-1K, to support research in unaligned multimodal and UAV-ground collaborative tracking. AI
IMPACT Provides a new resource for advancing the robustness of tracking systems in dynamic, real-world scenarios.
RANK_REASON The cluster describes a new academic paper introducing a benchmark dataset for a specific computer vision task. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
