This repository holds the Pytorch implementation of [ST-GCN-AltFormer:Gesture Recognition with Spatial-Temporal Alternating Transformer] Qing Pan, Jintao Zhu, Lingwei Zhang, Gangmin Ning, and Luping Fang.
We propose a Spatial-Temporal Alternating Transformer (AltFormer) method for hand gesture recognition. The key idea is that the present approaches have limitations in capturing the information conveyed in the synergistic actions of non-adjacent graph nodes, and their long-range dependencies. The code of training our approach for skeleton-based hand gesture recognition on the DHG-14/28 Dataset, the SHREC’17 Track Dataset and the LMDHG Dataset are provided in this repository.
This package has the following requirements:
Python 3.8
Pytorch v2.0.1
-
Download the DHG-14/28 Dataset , the SHREC’17 Track Dataset and LMDHG Dataset.
-
Run one of following commands.
python SHREC/ST_TS/train_sttran.py # on SHREC’17 Track Dataset
python LMDHG/ST_TS/LMDHG_sttran.py # on LMDHG Dataset
python DHG/ST_TS/DHG_sttran.py # on DHG Dataset
3. if you need weighting parameter(.pth), please download from Google Cloud Drive:https://drive.google.com/file/d/1BVXWKuMRgqca4v5DujPjHuCycqfp503D/view?usp=drive_link
4. Finally,run esemble.py or emsemble_LMDHG.py