Thermal soccer dataset - Tracking sports players

Context Simultaneous tracking of multiple people is still a very challenging computer vision problem. This is especially true for sports activities, for which people often wear similar uniforms, move quickly and erratically, and have close interactions with each other. This dataset is captured with thermal cameras, which enables easier segmentation and ensures privacy of people in public facilities, but at the same time we are left with no distinct appearance information to guide our tracking algorithms. Content This dataset contains four 30-seconds video sequences of eight people playing soccer in an indoor arena (court size 40*20 metres). The video is captured by thermal cameras of type AXIS Q1922 with a resolution of 640*480 pixels and 25 fps. The three images are stitched to one image of 1920*480 pixels. The videos are manually annotated for tracking. Acknowledgements Gade, R. & Moeslund, T.B.: Constrained multi-target tracking for team sports activities. IPSJ Transactions on Computer Vision and Applications (2018) 10: 2. https://doi.org/10.1186/s41074-017-0038-z