代码地址:
 https://github.com/facebookresearch/SlowFast/ 代码环境
 https://github.com/facebookresearch/SlowFast/blob/main/INSTALL.md
 
 行为识别模型
 Slowfast复现
 Python 版本要求3.8
 0.环境安装
 这里注意安装detectron2用如下命令
git clone https://github.com/facebookresearch/detectron2.git
python -m pip install -e detectron21.1构建slowfast
 下载好slowfast代码后进入到
cd SlowFast
python setup.py build develop1.2下载数据
 根据slowfast/DATASET.md中的路径下载
 Download the videos via the official [scripts](https://github.com/activitynet/ActivityNet/tree/master/Crawler/Kinetics). 2.运行测试脚本进行推理
 这里要先准备.pkl预训练模型,官方下载地址:https://github.com/facebookresearch/SlowFast/blob/main/MODEL_ZOO.md
 选择这个

 下载好后放入SlowFast文件夹下
 记得如果是cpu运行将yaml文件中的参数设置为NUM_GPUS: None
python tools/run_net.py --cfg demo/AVA/SLOWFAST_32x2_R101_50_50.yamlpython tools/run_net.py --cfg configs/Kinetics/C2D_8x8_R50.yaml NUM_GPUS 1 TRAIN.BATCH_SIZE 8 SOLVER.BASE_LR 0.0125 DATA.PATH_TO_DATA_DIR path_to_your_data_folder这里要指定path_to_your_data_folder数据集路径
 mmaction2
 https://github.com/open-mmlab/mmaction2
                










