[Project Page] [Code] [Arxiv] [Video] [Poster]
This is the official repository of our CVPR 2024 paper, "PartDistill: 3D Shape Part Segmentation by Vision-Language Model Distillation".
- The prapared data can be downloaded here. The content details are explained in data_format.md. After downloading, please extract the zip file.
- Point-M2AE checkpoint can be download here
To train the model, please use script bellow
CUDA_VISIBLE_DEVICES=<gpu_idx> python train.py \
--data_path<dir_path_storing_preprocessed_data> \
--ckpt_dir=<dir_to_store_ckpt> \
--backbone_path=<path_to_pointm2ae_ckpt> \
--lr=0.003 \
--batch_size=16 \
--category chair knife \
--n_epoch=25 \
--exp_suffix=exp1
For example:
CUDA_VISIBLE_DEVICES=8 python train.py \
--data_path=/disk2/aumam/dataset/shapenet/shapenet_preprocessed_partdistill \
--ckpt_dir=/disk2/aumam/dev/multimodal_distillation/checkpoints \
--backbone_path=/disk2/aumam/dev/multimodal_distillation/checkpoints/point_m2ae_pre-train.pth \
--lr=0.003 \
--batch_size=16 \
--category chair knife \
--n_epoch=25 \
--exp_suffix=exp1
pip install --upgrade https://github.com/unlimblue/KNN_CUDA/releases/download/0.2/KNN_CUDA-0.2-py3-none-any.whl
@inproceedings{umam2023partdistill,
title = {PartDistill: 3D Shape Part Segmentation by Vision-Language Model Distillation},
author = {Umam, Ardian and Yang, Cheng-Kun and Chen, Min-Hung and Chuang, Jen-Hui and Lin, Yen-Yu},
booktitle = {IEEE/CVF International Conference on Computer Vision (CVPR)},
year = {2024},
}