Introduction

Released code for the paper: a single-stage monocular 3D object detection model is proposed. An instance-segmentation head is integrated into the model training, which allows the model to be aware of the visible shape of a target object.

Installation

The code is tested with python==3.7, torch==1.4.0, torchvision==0.5.0.

After python and pytorch are installed on your machine, You can run the following commands to prepare the environment:

pip install -r requirements.txt 
cd mmdet3d/ops/DCNv2
. make.sh 
cd ../../..
pip install -v -e .

Data Preparation

Please download the KITTI dataset and organize the data as follows

data/kitti/     
        |training/
            |calib/
            |image_2/
            |label/
            |ImageSets/
        |testing/
            |calib/
            |image_2/
            |ImageSets/

Training & Evaluation

training:

python -m torch.distributed.launch --nproc_per_node=<number of gpus> mono3d/train.py <config file> --no-validate --launcher=pytorch

evaluation:

python mono3d/test.py <config file> <checkpoint file> --eval bbox

Model Zoo

Config File	Model	AP(3D)
configs_shapeaware/shapeaware_quantizationFloor_albu_SAFLw2NoRepair_CS.py	Google Drive	18.39

Acknowledgements