Sparse Coding Library for PyTorch
Installation
To install the package, clone it to a local folder, then run pip install -e . from the folder to install the package in edit mode (so changes to the code are used).
Overview
Main sparse coding file is found at: scripts/train_conv3d_sparse_model.py
Sparse coding command example: python scripts/train_conv3d_sparse_model.py
Main classification model is found at: scripts/train_classifier.py
Classification command example: python data_classifiers/small_data_classifier.py --checkpoint sparse.pt
Run on a single Yolo clip with: python scripts/run_on_single_image.py Notes: Need to have classifier.pt and sparse.pt (checkpoints) in root Videos go in a directory called input_videos