GraspClutter6D API

Overview

This repository contains dataset toolkit for the paper "GraspClutter6D: A Large-scale Real-world Dataset for Robust Perception and Grasping in Cluttered Scenes." (RA-L 2025)

[ArXiv] [Website] [Dataset] [Video]

• Load and manipulate 6D grasp pose annotations
• Perform grasp evaluation for benchmarking
• Maintain compatibility with graspnetAPI from GraspNet-1B

Dataset

The GraspClutter6D Dataset is available through Hugging Face

Installation

Option 1: Install from source (Recommended for latest version)

# Create and activate conda environment
conda create -n gc6d python=3.8
conda activate gc6d

# Clone repository and install in development mode
git clone https://github.com/SeungBack/graspclutter6dAPI.git
cd graspclutter6dAPI
pip install -e .

Option 2: Install via pip

pip install graspclutter6dAPI

Environment Setup

Before running the examples, set the environment variable to point to your dataset location:

export GC6D_ROOT='/path/to/graspclutter6d'

Usage Examples

Validate Dataset Integrity

Check if the downloaded dataset is complete and properly structured:

python examples/exam_check_data.py

Load Grasp Annotations

Load and process grasp annotations from the GraspClutter6D dataset:

python examples/exam_loadGrasp.py

Visualize Grasp Annotations

Generate visualizations of grasp poses and objects:

python examples/exam_vis.py

Acknowledgments

This repository is built upon the graspnetAPI. We thank the authors for sharing their code.