wish20241105

# Parameter is Not All You Need Official implementation of ['Parameter is Not All You Need: Starting from Non-Parametric Networks for 3D Point Cloud Analysis'](https://arxiv.org/pdf/2303.08134.pdf). The paper has been accepted by **CVPR 2023** 🔥. [](https://paperswithcode.com/sota/training-free-3d-point-cloud-classification?p=parameter-is-not-all-you-need-starting-from) [](https://paperswithcode.com/sota/training-free-3d-point-cloud-classification-1?p=parameter-is-not-all-you-need-starting-from) [](https://paperswithcode.com/sota/training-free-3d-part-segmentation-on?p=parameter-is-not-all-you-need-starting-from) ## News * **Seg-NN** has been accepted as ***CVPR 2024 Highlight Paper*** 🔥! * We release [Seg-NN](https://arxiv.org/pdf/2404.04050.pdf) and [code](https://github.com/yangyangyang127/Seg-NN), which adapts Point-NN & Point-PN into 3D scene segmentation tasks 🔥. * For the first time, we conduct 3D analysis entirely requiring $\color{darkorange}{No\ Parameter\ or\ Training\.}$ 💥 * The code of Point-PN has been released 📌. * The code of Point-NN for shape classification and part segmentation has been released. ## Introduction We present a **N**on-parametric **N**etwork for 3D point cloud analysis, **Point-NN**, which consists of purely non-learnable components. Surprisingly, requiring no parameters or training, it performs well on various 3D tasks, and even surpasses existing fully trained models. Starting from this basic non-parametric model, we propose two extensions. First, Point-NN can serve as a base architectural framework to construct **P**arametric **N**etworks, **Point-PN**, which exhibits superior performance with simple linear layers. Second, Point-NN can be regarded as a plug-and-play module to enhance the already trained 3D models during inference by complementary knowledge. ## Requirements ### Installation Create a conda environment and install dependencies: ```bash git clone https://github.com/ZrrSkywalker/Point-NN.git cd Point-NN conda create -n pointnn python=3.7 conda activate pointnn # Install the according versions of torch and torchvision conda install pytorch torchvision cudatoolkit pip install -r requirements.txt pip install pointnet2_ops_lib/. ``` ### Dataset Please download the following datasets: [ModelNet40](https://shapenet.cs.stanford.edu/media/modelnet40_ply_hdf5_2048.zip), [ScanObjectNN](https://hkust-vgd.ust.hk/scanobjectnn/h5_files.zip), and [ShapeNetPart](https://shapenet.cs.stanford.edu/media/shapenetcore_partanno_segmentation_benchmark_v0_normal.zip). Then, create a `data/` folder and organize the datasets as ``` data/ |–– h5_files/ |–– modelnet40_ply_hdf5_2048/ |–– shapenetcore_partanno_segmentation_benchmark_v0_normal/ ``` ## Point-NN --- Very Quick Implementation 🚀 ### Shape Classification Due to the training-free manner, the preparation and inference of Point-NN only take **2 minutes**. For ModelNet40 dataset, just run: ```bash python run_nn_cls.py --dataset mn40 ``` For ScanObjectNN dataset, just run: ```bash python run_nn_cls.py --dataset scan --split 1 ``` Please indicate the splits at `--split` by `1,2,3` for OBJ-BG, OBJ-ONLY, and PB-T50-RS, respectively. ### Part Segmentation For ShapeNetPart, Point-NN takes **7 minutes** to achieve 71.5% mIOU (70.4% in the paper), just run: ```bash python run_nn_seg.py ``` You can increase the point number `--points` and k-NN neighbors `--k` into `2048` and `128`, which further acheives **74%** with 18 minutes. ## Point-PN ### Shape Classification Point-PN is the parametric version of Point-NN with efficient parameters and simple 3D operators. For ModelNet40 dataset, just run: ```bash python run_pn_mn40.py --msg

``` For ScanObjectNN dataset, just run: ```bash python run_pn_scan.py --split 1 --msg

``` Please indicate the splits at `--split` by `1,2,3` for OBJ-BG, OBJ-ONLY, and PB-T50-RS, respectively. ## Citation ```bash @article{zhang2023parameter, title={Parameter is not all you need: Starting from non-parametric networks for 3d point cloud analysis}, author={Zhang, Renrui and Wang, Liuhui and Wang, Yali and Gao, Peng and Li, Hongsheng and Shi, Jianbo}, journal={arXiv preprint arXiv:2303.08134}, year={2023} } ``` ## Contact If you have any question about this project, please feel free to contact zhangrenrui@pjlab.org.cn.