- 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 <=1.8
- TorchVision >= 0.7.0 <=0.9.0
- mmcv-full (according to PyTorch/CUDA version)
pip install -r requirements.txt
git clone https://github.com/voldemortX/pytorch-auto-drive.git
cd pytorch-auto-drive
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.