BadmintonGRF: A Multimodal Dataset and Benchmark for Markerless Ground Reaction Force Estimation in Badminton

BadmintonGRF (v1.0) is a multi-modal dataset designed for estimating ground reaction forces (GRF) from badminton videos. This release includes processed impact segments, multi-view pose data, GRF signals/labels, synchronization metadata, and LOSO splits for 10 subjects. Raw videos and motion capture data are available via separate access. Due to the platform's 50GB storage limit, uploading the complete Tier 2 (controlled-access) dataset is impractical. For reviewers' convenience, we have provided a single-subject Tier 2 demo sample in this record. The full dataset will be hosted on an external cloud drive.