Skip to content

Latest commit

 

History

History
45 lines (37 loc) · 1.43 KB

ENV.md

File metadata and controls

45 lines (37 loc) · 1.43 KB

Environment

You need to create an environment that meets the following dependencies. The following are two examples of environments. The versions included in the parenthesis are tested. Other versions may also work but are not tested.

A tested local PC with NVIDIA TITAN Xp GPUs (2*12GB, driver 396.26):

  • Python (3.7.9)
  • Numpy (1.19.2)
  • PyTorch (1.7.1, GPU required)
  • Scipy (1.5.4)
  • Matplotlib (3.3.3)
  • OpenCV-python (4.4.0.46)
  • CUDA (9.2.88)

Another tested computing node with NVIDIA RTX 3090 GPUs (8*24 GB, driver 470.86):

  • Python (3.7.9)
  • Numpy (1.20.3)
  • PyTorch (1.9.0, GPU required)
  • Scipy (1.6.2)
  • Matplotlib (3.4.2)
  • OpenCV (3.4.2)
  • CUDA (11.4)

For more details of the environment, you can refer to the spec-file.txt.

The recommended environment manager is Anaconda, which can create an environment using this provided spec-list. For debugging using an IDE, the recommended IDE is Spyder which you can get by

conda install spyder

Compile and install

Compile and install the extension modules at your project directory ${SNVC_DIR} with:

python setup.py develop

Compile the KITTI evaluator (recommended)

If you need to quantitatively evaluate the model predictions for KITTI, you need to compile this evaluation tool.

cd ${SNVC_DIR}/tools/kitti-eval

Compile the source code

g++ -o evaluate_object_3d_offline evaluate_object_3d_offline.cpp -O3