Skip to content

Latest commit

 

History

History
56 lines (40 loc) · 1.98 KB

README.md

File metadata and controls

56 lines (40 loc) · 1.98 KB

modelnet40_experiments

This repository contains experiments conducted on the ModelNet40 dataset. The ModelNet40 dataset is a widely used benchmark dataset for 3D classification tasks.

Dataset

The ModelNet40 dataset consists of 12,311 CAD models from 40 object categories, with approximately 300 instances per category. The dataset is divided into two sets: a training set of 9,843 models and a test set of 2,468 models. The original CAD models are provided in the .off format.

The experiments are conducted on the processed data in .hdf5 format. Each processed object contains 2048 points sampled from the original CAD models.

Install Dependencies

pip install -r requirements.txt

Experiments

The experiments are conducted on the following models:

Training models by running the corresponding scripts in the code folder. For example, to train the DGCNN model, run the following command:

python code/train_dgcnn.py

Results

The table below presents the classification accuracy of the models on the ModelNet40 dataset (1024 points, trained on a single Nvidia RTX 3090 GPU). Some of the following results are the best outcomes obtained after hyperparameter sweepings.

Model input Overall Accuracy sweep
DGCNN xyz 91.7%
PointNet xyz 90.7%
PointNet2SSG xyz 90.7%
PointNet2MSG xyz 92.1%
PointNext xyz 91.3%

You can reproduce the results by running the corresponding scripts in the code folder with default configurations. For example, to train the PointNet model, run the following command

python code/train_pointnet.py