git clone https://github.com/voldemortX/pytorch-auto-drive.git
cd pytorch-auto-drive
- Linux (recommended) or Windows (not fully tested, could have problems)
- Python >= 3.6
- CUDA >= 9.2 (for CUDA version < 9.2, the code is tested only with PyTorch 1.3 & CUDA 9.0 & CuDNN 7.6.0)
- PyTorch >= 1.6 (2.x are not tested)
- TorchVision >= 0.7.0
- mmcv-full >= 1.3.5 (according to PyTorch/CUDA version)
- Other pip dependencies:
pip install -r requirements.txt
The default Conda env (step-by-step):
conda create -n pad python=3.6
conda activate pad
conda install pytorch==1.6.0 torchvision==0.7.0 cudatoolkit=10.2 -c pytorch
pip install mmcv-full==1.3.5 -f https://download.openmmlab.com/mmcv/dist/cu102/torch1.6.0/index.html
pip install -r requirements.txt
chmod 777 *.sh tools/shells/*.sh
mkdir output
Improve training speed with Pillow-SIMD (optional, advanced):
pip uninstall pillow
CC="cc -mavx2" pip install -U --force-reinstall pillow-simd
Note that you need to use ToTensor transform as late as possible for this speedup.
tensorboard --logdir=<path to tb_logs>
<path to tb_logs>
is usually ./checkpoints/tb_logs
if you did not customized save_dir
in config file.