seantangth/WACV2027-RetailAction-Challenge-weights

RetailAction Challenge (PRAW @ WACV 2027) — model weights, team seantangth

Checkpoints used by the inference code at https://github.com/seantangth/WACV2027-RetailAction-Challenge (Docker, single command). They reproduce our final submission (public leaderboard 0.9955). See MANIFEST.txt for the role of each file and SHA256SUMS.txt for checksums.

Attribution. These models were trained on the RetailAction dataset, authored by Standard Cognition (Standard AI) and used under the Standard AI RetailAction dataset license (https://huggingface.co/datasets/standard-cognition/RetailAction). Paper: Mazzini et al., "RetailAction: Dataset for Multi-View Spatio-Temporal Localization of Human-Object Interactions in Retail", ICCV 2025 Retail Vision Workshop. The dataset itself is not included.