Data

Tools

Indexes

Applications

Experiments

Open Source Science

https://github.com/asifalfaisal/3d_object_classification_pointclouds_gnn

https://github.com/asifalfaisal/3d_object_classification_pointclouds_gnn

Science Score: 10.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ○
    CITATION.cff file
  • ○
    codemeta.json file
  • ○
    .zenodo.json file
  • ○
    DOI references
  • ✓
    Academic publication links
    Links to: arxiv.org
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (6.3%) to scientific vocabulary
Last synced: 6 months ago · JSON representation

Repository

Basic Info
  • Host: GitHub
  • Owner: AsifAlFaisal
  • License: mit
  • Language: Python
  • Default Branch: main
  • Size: 3.23 MB
Statistics
  • Stars: 1
  • Watchers: 2
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Created about 4 years ago · Last pushed about 4 years ago

https://github.com/AsifAlFaisal/3d_object_classification_pointclouds_gnn/blob/main/ 

# 3D Object Classification from Point Cloud using Graph Neural Net
**Dataset**: [ModelNet10 benchmark data](https://people.csail.mit.edu/khosla/papers/cvpr2015_wu.pdf)  a large-scale 3D CAD model dataset containing 10 classes.
- **Classes:** bathtub, bed, chair, desk, dresser, monitor, night_stand, sofa, table, toilet

### Brief Description:
- First step was to transform 3D mesh objects to pointcloud objects.
- From these point clouds, graphs were constructed based on nearest neighboring nodes.
- After previous two steps, a rotation invariant framework [**Point Pair Feature Net(PPFNet)**](https://arxiv.org/abs/1802.02669) has been implemented and modified to handle graph point cloud data.

### Test Result:
| Accuracy  | Balanced Accuracy |
| ------ | ------ |
| 0.76  | 0.74 |

### Libraries and Frameworks
- For model building: **PyTorch** & **PyTorch Geometric**
- For hyperparameter tuning: **Optuna**
- For tracking model performance: **MLflow**
- Object Visualization: **Plotly**

### Sample 3D Objects:
Sample representation of 3D Mesh Objects and their respective Point Clouds are given below:

![alt text](interactive_sample_visualization/static/sample1.PNG)

![alt text](interactive_sample_visualization/static/sample2.PNG)

Owner

  • Name: Asif Al Faisal
  • Login: AsifAlFaisal
  • Kind: user

GitHub Events

Total
Last Year