JATAYU: A Large-scale Indian UAV Dataset for Vehicle Detection and Tracking
Abhijnan Maji, Kummara Preetham, Indrajit Ghosh · 2024
Advances in Unmanned Aerial Vehicle (UAV) technology require extensive vehicular datasets for a wide range of applications within Intelligent Transportation Systems. However, the existing datasets partially cover the real-world complexities and the representation of a wide variety of vehicle types prevalent in the traffic streams of developing countries. The "Joint Aerial dataset for deTection and trAcking of vehicles for road safetY and sUrveillance (JATAYU)" is introduced in this study to address these shortcomings. The dataset comprises varied vehicle classes across different Indian locales. The dataset categorizes vehicle types according to the Indian Highway Capacity Manual (Indo-HCM). The utility of this dataset spans the Indian subcontinent and Southeast Asian countries due to the presence of vehicles like auto-rickshaws, e-rickshaws, motorcycles, cycle-rickshaws, and tractor-trolleys in these regions. JATAYU contains 215,000 labeled objects with nearly 30,000 annotated images in ten vehicle classes. It includes daytime, night-time, and partially obscured vehicular aerial images.