Repository files navigation

Clone this repository:

git clone https://github.com/Zengyi-Qin/Weakly-Supervised-3D-Object-Detection.git

Enter the main folder and run installation:

pip install -r requirements.txt

Download thedemo data to the main folder and rununzip vs3d_demo.zip. Readers can try out the quick demo with Jupyter Notebook:

cd corejupyter notebook demo.ipynb

Download theKitti Object Detection Dataset (image,calib andlabel) and place them intodata/kitti. Download the ground planes and front-view XYZ maps fromhere and rununzip vs3d_train.zip. Download the pretrained teacher network fromhere and rununzip vs3d_pretrained.zip. The data folder should be in the following structure:

├── data│   ├── demo│   └── kitti│       └── training│           ├── calib│           ├── image_2│           ├── label_2│           ├── sphere│           ├── planes│           └── velodyne│       ├── train.txt│       └── val.txt│   └── pretrained│       ├── student│       └── teacher

Thesphere folder contains the front-view XYZ maps converted fromvelodyne point clouds using the script in./preprocess/sphere_map.py. After data preparation, readers can train VS3D from scratch by running:

cd corepython main.py --mode train --gpu GPU_ID

The models are saved in./core/runs/weights during training. Reader can refer to./core/main.py for other options in training.

Readers can run the inference on KITTI validation set by running:

cd corepython main.py --mode evaluate --gpu GPU_ID --student_model SAVED_MODEL

Readers can also directly use the pretrained model for inference by passing--student_model ../data/pretrained/student/model_lidar_158000. Predicted 3D bounding boxes are saved in./output/bbox in KITTI format.

@article{qin2020vs3d,   title={Weakly Supervised 3D Object Detection from Point Clouds},   author={Zengyi Qin and Jinglu Wang and Yan Lu},  journal={ACM Multimedia},  year={2020}}