pininduwk / pointnet-for-3d-object-classification-pytorch-

PointNet for 3D Object Classification (PyTorch)

This repository contains an implementation of PointNet for 3D object classification in PyTorch, based on the paper "PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation" by Charles R. Qi et al.

Dataset

The model is trained and tested on the ModelNet10 dataset, which contains 10 categories of objects (bed, chair, dresser, etc.) with 4899 objects in total.

Requirements

The code requires the following libraries:

• PyTorch
• NumPy
• open3d
• scikit-learn
• matplotlib
• plotly
• tqdm

These can be installed using pip:

pip install torch numpy open3d scikit-learn matplotlib plotly tqdm

Training

To train the model, run:

python train.py --data_path /path/to/dataset --batch_size 32 --num_epochs 50 --learning_rate 0.001

You can adjust the batch size, number of epochs, and learning rate as needed.

Evaluation

To evaluate the trained model on the test set, run:

python test.py --data_path /path/to/dataset --model_path /path/to/model

Replace /path/to/dataset with the path to the ModelNet10 dataset and /path/to/model with the path to the saved model checkpoint.

Results

The model achieves an accuracy of 90.4% on the test set.

References

• Qi, Charles R., et al. "PointNet: Deep Learning on Point Sets for 3D Classification and Segmentation." Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017.
• ModelNet10 Dataset: https://modelnet.cs.princeton.edu/