Fine-Grained Action Detection with RGB and Pose Information using Two Stream Convolutional Networks
This work addresses action detection in sports analytics, specifically for table tennis, but is incremental as it builds on existing two-stream and attention mechanisms.
The paper tackles fine-grained action detection for table tennis strokes by proposing a two-stream network using RGB and pose information, achieving 87.3% accuracy in stroke classification but not outperforming the baseline in detection with an IoU of 0.349 and mAP of 0.110.
As participants of the MediaEval 2022 Sport Task, we propose a two-stream network approach for the classification and detection of table tennis strokes. Each stream is a succession of 3D Convolutional Neural Network (CNN) blocks using attention mechanisms. Each stream processes different 4D inputs. Our method utilizes raw RGB data and pose information computed from MMPose toolbox. The pose information is treated as an image by applying the pose either on a black background or on the original RGB frame it has been computed from. Best performance is obtained by feeding raw RGB data to one stream, Pose + RGB (PRGB) information to the other stream and applying late fusion on the features. The approaches were evaluated on the provided TTStroke-21 data sets. We can report an improvement in stroke classification, reaching 87.3% of accuracy, while the detection does not outperform the baseline but still reaches an IoU of 0.349 and mAP of 0.110.