%0 Conference Proceedings
%A Martin, Pierre-Etienne
%+ Department of Comparative Cultural Psychology, Max Planck Institute for Evolutionary Anthropology, Max Planck Society
%T Baseline method for the Sport Task of MediaEval 2022 with 3D CNNs using attention mechanisms : 
%G eng
%U https://hdl.handle.net/21.11116/0000-000E-162D-8
%R 10.48550/arXiv.2302.02752
%D 2023
%8 06.02.2023
%Z Review method: peer-reviewed
%B MediaEval 2022: Multimedia Evaluation Workshop
%Z date of event: 2023-01-13 - 2023-01-15
%C  Bergen, Norway
%9 invited
%X This paper presents the baseline method proposed for the Sports Video task part of the MediaEval 2022 benchmark. This task proposes two subtasks: stroke classification from trimmed videos, and stroke detection from untrimmed videos. This baseline addresses both subtasks. We propose two types of 3D-CNN architectures to solve the two subtasks. Both 3D-CNNs use Spatio-temporal convolutions and attention mechanisms. The architectures and the training process are tailored to solve the addressed subtask. This baseline method is shared publicly online to help the participants in their investigation and alleviate eventually some aspects of the task such as video processing, training method, evaluation and submission routine. The baseline method reaches 86.4% of accuracy with our v2 model for the classification subtask. For the detection subtask, the baseline reaches a mAP of 0.131 and IoU of 0.515 with our v1 model.
%K Attention mechanisms, Baseline methods, Spatio-temporal, Sport video, Stroke detection, Subtask, Training methods, Training process, Video detection, Video processing
%B CEUR Workshop Proceedings: MediaEval 2022 : Multimedia Evaluation Workshop
%V 3583
%Z sequence number: 195253
%@ 1613-0073