前言
昨天有人给我发co-tracker开源,秉持着一贯好奇与好学的心态。尝试自己搭建环境跑跑demo看看效果。
git链接:https://github.com/facebookresearch/co-tracker
本机环境:ubuntu 18.04 cuda 11.7
一、Anaconda安装
推荐参考链接:https://blog.csdn.net/chloexxx/article/details/108736074
1、anaconda安装
wget https://repo.anaconda.com/archive/Anaconda3-2019.07-Linux-x86_64.sh
安装完成后:
2、验证anaconda是否安装成功
anaconda
出现下图说明安装成功
3、创建新的虚拟环境
以我的为例:
conda create -n Co-Trakcer python=3.7
4、激活虚拟环境:
conda activate Co-Trakcer
二、安装PyTorch and TorchVision with CUDA support
根据链接https://pytorch.org/get-started/locally/
conda install pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia
由于网络波动有时会超时,多试几次
三、根据提示安装
1、安装CoTracker及其依赖项
git clone https://github.com/facebookresearch/co-tracker
cd co-tracker
pip install -e .
pip install opencv-python einops timm matplotlib moviepy flow_vis
2、下载模型权重
mkdir checkpoints
cd checkpoints
wget https://dl.fbaipublicfiles.com/cotracker/cotracker_stride_4_wind_8.pth
wget https://dl.fbaipublicfiles.com/cotracker/cotracker_stride_4_wind_12.pth
wget https://dl.fbaipublicfiles.com/cotracker/cotracker_stride_8_wind_16.pth
cd ..
3、运行demo
如果提示缺少库,安装相应的库就行
python demo.py --grid_size 10
4、等待几秒钟计算
计算完成,提示保存了video