CompletionFormer 点云补全 学习笔记

目录

依赖项

数据集

KITTI DC

NYUv2 提供深度图

模型

---

依赖项

conda create -n completionformer python=3.8
conda activate completionformer
# For CUDA Version == 11.3
pip install torch==1.10.1+cu113 torchvision==0.11.2+cu113 torchaudio==0.10.1+cu113
pip install mmcv-full==1.4.4 mmsegmentation==0.22.1  
pip install timm tqdm thop tensorboardX opencv-python ipdb h5py ipython Pillow==9.5.0 

dcv v2

数据集

KITTI DC

depth velodyne 深度测速仪

NYUv2 提供深度图

$ cd PATH_TO_DOWNLOAD
$ wget http://datasets.lids.mit.edu/sparse-to-dense/data/nyudepthv2.tar.gz
$ tar -xvf nyudepthv2.tar.gz

After that, you will get a data structure as follows:

nyudepthv2
├── train
│    ├── basement_0001a
│    │    ├── 00001.h5
│    │    └── ...
│    ├── basement_0001b
│    │    ├── 00001.h5
│    │    └── ...
│    └── ...
└── val└── official├── 00001.h5└── ...

Note that the original full NYUv2 dataset is available at the official website.

After preparing the dataset, you should generate a json file containing paths to individual images.

$ cd THIS_PROJECT_ROOT/utils
$ python generate_json_NYUDepthV2.py --path_root PATH_TO_NYUv2

模型

318M