The pipeline to generate ST-GCN visual features for SMIE.
After splitting the data of different unseen classes using split.py, utilize gen_feature.py to obtain the corresponding visual features for ST-GCN.
Modify gen_config.py to select a different split_id for different datasets and customize the storage location.
$ python gen_feature.py
We also provide model weights for ST-GCN corresponding to the split_id in SMIE: Model Weights
code:smie