Task
Object Detection
Models trained on UVH-26 deliver up to 31.5% higher mAP than COCO-pretrained baselines, demonstrating significant gains in real-world performance for Indian traffic scenarios.
Models trained on UVH-26 deliver up to 31.5% higher mAP than COCO-pretrained baselines, demonstrating significant gains in real-world performance for Indian traffic scenarios.