This is the pytorch implementation of our paper.
PyTorch 1.4
PyTorch-Encoding v1.4.0
Compile the C++ and CUDA code:
cd exts
python setup.py install
Please download KITTI depth completion dataset. The structure of data directory:
└── datas
└── kitti
├── data_depth_annotated
│ ├── train
│ └── val
├── data_depth_velodyne
│ ├── train
│ └── val
├── raw
│ ├── 2011_09_26
│ ├── 2011_09_28
│ ├── 2011_09_29
│ ├── 2011_09_30
│ └── 2011_10_03
├── test_depth_completion_anonymous
│ ├── image
│ ├── intrinsics
│ └── velodyne_raw
└── val_selection_cropped
├── groundtruth_depth
├── image
├── intrinsics
└── velodyne_raw
The config of different settings:
Compared to GN, GNS uses fewer parameters to generate the guided kernels, but achieves slightly better results.
You can directly download the trained model and put it in checkpoints:
You can also train by yourself:
python train.py
Pay attention to the settings in the config file (e.g. gpu id).
With the trained model, you can test and save depth images.
python test.py
If you find this work useful in your research, please consider citing:
@article{guidenet,
title={Learning guided convolutional network for depth completion},
author={Tang, Jie and Tian, Fei-Peng and Feng, Wei and Li, Jian and Tan, Ping},
journal={IEEE Transactions on Image Processing},
volume={30},
pages={1116--1129},
year={2020},
publisher={IEEE}
}