
Python 3.6+
PyTorch 1.3+
CUDA 9.2+ (If you build PyTorch from source, CUDA 9.0 is also compatible)
GCC 5+
mmcv 1.1.1+
Numpy
创建虚拟环境
conda create -n open-mmlab python=3.7 -y
conda activate open-mmlab
安装对应自己cuda 版本的pytorch,我的cude版本是10.1所以我使用的是:
conda install pytorch==1.7.1 torchvision==0.8.2 torchaudio==0.7.2 cudatoolkit=10.1 -c pytorch
(这里可以参考连接:https://pytorch.org/get-started/previous-versions
安装MMAction2
安装MIM
pip install git+https://github.com/open-mmlab/mim.git
mim install mmaction2 -f https://github.com/open-mmlab/mmaction2.git
安装 mmcv-full(注意这里我们之前的pytorch版本是1.7.1但是因为安装这个只对1.x.0有效,所以我使用1.7.0)
pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu101/torch1.7.0/index.html
pip install mmcv
克隆MMAction2 repository.
git clone https://github.com/open-mmlab/mmaction2.git
cd mmaction2
安装build requirements 以及安装MMAction2.
pip install -r requirements/build.txt
pip install -v -e .
验证是否安装成功
import torch
from mmaction.apis import init_recognizer, inference_recognizer
config_file = 'configs/recognition/tsn/tsn_r50_video_inference_1x1x3_100e_kinetics400_rgb.py'
device = 'cuda:0' # 或 'cpu'
device = torch.device(device)
model = init_recognizer(config_file, device=device)
inference_recognizer(model, 'demo/demo.mp4')
我这里是直接在我的终端执行的
测试结果为:
[(270, 31.022188), (230, 27.026155), (117, 24.818554), (318, 24.535355), (282, 24.215961)]
欢迎分享,转载请注明来源:内存溢出
微信扫一扫
支付宝扫一扫
评论列表(0条)