TY - GEN
T1 - SoccerNet-v2: A Dataset and Benchmarks for Holistic Understanding of Broadcast Soccer Videos
AU - Deliège, Adrien
AU - Cioppa, Anthony
AU - Giancola, Silvio
AU - Seikavandi, Meisam J.
AU - Dueholm, Jacob V.
AU - Nasrollahi, Kamal
AU - Ghanem, Bernard
AU - Moeslund, Thomas B.
AU - Droogenbroeck, Marc Van
N1 - KAUST Repository Item: Exported on 2021-09-03
PY - 2021
Y1 - 2021
N2 - Understanding broadcast videos is a challenging task in computer vision, as it requires generic reasoning capabilities to appreciate the content offered by the video editing. In this work, we propose SoccerNet-v2, a novel large-scale corpus of manual annotations for the SoccerNet [24] video dataset, along with open challenges to encourage more research in soccer understanding and broadcast production. Specifically, we release around 300k annotations within SoccerNet’s 500 untrimmed broadcast soccer videos. We extend current tasks in the realm of soccer to include action spotting, camera shot segmentation with boundary detection, and we define a novel replay grounding task. For each task, we provide and discuss benchmark results, reproducible with our open-source adapted implementations of the most relevant works in the field. SoccerNet-v2 is presented to the broader research community to help push computer vision closer to automatic solutions for more general video understanding and production purposes.
AB - Understanding broadcast videos is a challenging task in computer vision, as it requires generic reasoning capabilities to appreciate the content offered by the video editing. In this work, we propose SoccerNet-v2, a novel large-scale corpus of manual annotations for the SoccerNet [24] video dataset, along with open challenges to encourage more research in soccer understanding and broadcast production. Specifically, we release around 300k annotations within SoccerNet’s 500 untrimmed broadcast soccer videos. We extend current tasks in the realm of soccer to include action spotting, camera shot segmentation with boundary detection, and we define a novel replay grounding task. For each task, we provide and discuss benchmark results, reproducible with our open-source adapted implementations of the most relevant works in the field. SoccerNet-v2 is presented to the broader research community to help push computer vision closer to automatic solutions for more general video understanding and production purposes.
UR - http://hdl.handle.net/10754/670907
UR - https://ieeexplore.ieee.org/document/9523091/
U2 - 10.1109/CVPRW53098.2021.00508
DO - 10.1109/CVPRW53098.2021.00508
M3 - Conference contribution
SN - 978-1-6654-4900-7
BT - 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
PB - IEEE
ER -