excitohe / monoex

MonoEX

MonoEX: Excito Private Monocular 3D Object Detection Lib.

Work in progress.

Installation

python setup.py develop

Dataset

We train and test our model on official KITTI 3D Object Dataset. Please first download the dataset and organize it as following structure:

kitti
│──ImageSets
│──training
│    ├──calib
│    ├──label_2 
│    └──image_2
└──testing
     ├──calib
     └──image_2

Then modify the paths in config/paths_catalog.py according to your data path.

Benchmark

Supported methods and backbones are shown in the below table.

Support backbones:

• DLA

Support methods

• MonoFlex (CVPR'2021)
• SMOKE (CVPR'2020)

Acknowledgement

The codebase is heavily borrowed from following projects:

• MonoFlex
• SMOKE
• mmcv

Thanks for their contribution.