INSTALL.md

Installation

The code was tested on Ubuntu 16.04, with Anaconda Python 3.6, CUDA 10.0, and PyTorch v1.0. It should be compatible with PyTorch <=1.4 and python >=0.4 (you will need to switch DCNv2 version for PyTorch <1.0). After installing Anaconda:

0. [Optional but highly recommended] create a new conda environment.

   conda create --name CenterTrack python=3.6
   

   And activate the environment.

   conda activate CenterTrack
   

1. Install PyTorch:

   conda install pytorch torchvision -c pytorch
   

2. Install COCOAPI:

   pip install cython; pip install -U 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'
   

3. Clone this repo:

   CenterTrack_ROOT=/path/to/clone/CenterTrack
   git clone --recursive https://github.com/xingyizhou/CenterTrack $CenterTrack_ROOT
   

   You can manually install the submodules if you forget --recursive.

4. Install the requirements

   pip install -r requirements.txt
   

5. Compile deformable convolutional (from DCNv2).

   cd $CenterTrack_ROOT/src/lib/model/networks/
   # git clone https://github.com/CharlesShang/DCNv2/ # clone if it is not automatically downloaded by `--recursive`.
   cd DCNv2
   ./make.sh
   

6. Download pertained models for monocular 3D tracking, 80-category tracking, or pose tracking and move them to $CenterTrack_ROOT/models/. More models can be found in Model zoo.