DGCNN Paddle Version

Original paper: Dynamic Graph CNN for Learning on Point Clouds

Dataset: ShapeNet

Original Pytorch Implementation: DGCNN-Pytorch

Introduction

This repository contains a Paddle implementation of DGCNN in the paper "Dynamic Graph CNN for Learning on Point Clouds". Our code is based on PaddlePaddle 2.2.0, so you need to install paddlepaddle first.

We provide our trained model and put it in the folder 'pretrained'.

Getting Started

• Run the training script:

python main.py --exp_name=dgcnn_1024 --model=dgcnn --num_points=1024 --k=20 --use_sgd=True

python main.py --exp_name=dgcnn_2048 --model=dgcnn --num_points=2048 --k=40 --use_sgd=True

• Run the evaluation script after training finished:

python main.py --exp_name=dgcnn_1024_eval --model=dgcnn --num_points=1024 --k=20 --use_sgd=True --eval=True --model_path=checkpoints/dgcnn_1024/models/model.pdparams

python main.py --exp_name=dgcnn_2048_eval --model=dgcnn --num_points=2048 --k=40 --use_sgd=True --eval=True --model_path=checkpoints/dgcnn_2048/models/model.pdparams

• Run the evaluation script with pretrained models:

python main.py --exp_name=dgcnn_1024_eval --model=dgcnn --num_points=1024 --k=20 --use_sgd=True --eval=True --model_path=pretrained/model.pdparams