DFU: A Python repository from YufeiWang777

Improving Depth Completion via Depth Feature Upsampling (CVPR 2024)

Environment

CUDA 12.0
CUDNN 8.5.0
torch 1.7.1
torchvision 0.8.0
pip install -r LRRU/requirements.txt
pip3 install opencv-python
pip3 install opencv-python-headless

Wandb

We used WANDB to visualize and track our experiments.

NVIDIA Apex

We used NVIDIA Apex for multi-GPU training as NLSPN.

Apex can be installed as follows:

$ cd PATH_TO_INSTALL
$ git clone https://github.com/NVIDIA/apex
$ cd apex
$ pip install -v --disable-pip-version-check --no-build-isolation --no-cache-dir ./

Dataset

KITTI Depth Completion (KITTI DC)

KITTI DC dataset is available at the KITTI DC Website and the data structure is:

.
├── depth_selection
│    ├── test_depth_completion_anonymous
│    │    ├── image
│    │    ├── intrinsics
│    │    └── velodyne_raw
│    ├── test_depth_prediction_anonymous
│    │    ├── image
│    │    └── intrinsics
│    └── val_selection_cropped
│        ├── groundtruth_depth
│        ├── image
│        ├── intrinsics
│        └── velodyne_raw
├── train
│    ├── 2011_09_26_drive_0001_sync
│    │    ├── image_02
│    │    │     └── data
│    │    ├── image_03
│    │    │     └── data
│    │    ├── oxts
│    │    │     └── data
│    │    └── proj_depth
│    │        ├── groundtruth
│    │        └── velodyne_raw
│    └── ...
└── val
    ├── 2011_09_26_drive_0002_sync
    └── ...

Usage

Training

$ sh train.sh

Testing

# download the pretrained model and add it to corresponding path.

$ sh val.sh

Pretrained models

Models on the KITTI validate dataset.

LRRU_with_Three_DFU

Acknowledgments

Thanks the ACs and the reviewers for their insightful comments, which are very helpful to improve our paper!

Citation

@inproceedings{wang2024improving,
  title={Improving Depth Completion via Depth Feature Upsampling},
  author={Wang, Yufei and Zhang, Ge and Wang, Shaoqian and Li, Bo and Liu, Qi and Hui, Le and Dai, Yuchao},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={21104--21113},
  year={2024}
}

YufeiWang777/DFU