The GOT-10k dataset contains more than 10,000 video segments of real-world moving objects and over 1.5 million manually labelled bounding boxes.
207 PAPERS • 2 BENCHMARKS