Action Recognition
Action Detection
- SlowFast (ICCV 2019) (Coming Soon)
Datasets
- Kinetics-400/600/700 (CVPR 2017)
- AVA (CVPR 2018) (Coming Soon)
Refer to MODELS.md for supported pre-trained models and benchmarks of SOTA models.
- python >= 3.6
- torch >= 1.9.0
- torchvision >= 0.10.0
Then, clone and install this repo with:
$ git clone https://github.com/sithu31296/video-classification
$ cd video-classification
$ pip install -e .
Use the following script to test the pre-trained model:
$ python tools/infer.py \
--source VIDEO_FILE_NAME
--model MODEL_NAME
--model_path PRETRAINED_MODEL_PATH
--num_classes DATASET_NUM_CLASSES
You will get the top-5 score similar to this:
Class Score (%)
------------- -----------
archery 91.49
throwing axe 0.11
slacklining 0.06
feeding fish 0.06
rock climbing 0.05
Follow the steps provided in DATASETS.md.
Coming Soon
Coming Soon
Coming Soon
@misc{li2022uniformer,
title={UniFormer: Unified Transformer for Efficient Spatiotemporal Representation Learning},
author={Kunchang Li and Yali Wang and Peng Gao and Guanglu Song and Yu Liu and Hongsheng Li and Yu Qiao},
year={2022},
eprint={2201.04676},
archivePrefix={arXiv},
primaryClass={cs.CV}
}