This project provides a Unity-based framework to create multi-modal datasets for machine learning tasks. With an easy-to-use setup, you can generate high-quality synthetic data from Unity simulations, including RGB images, object segmentation, normals, depth, and outlines.
The framework is built on top of the Unity ML-ImageSynthesis project and extends its functionality to meet various data generation requirements.
- Generate a wide range of modalities, including RGB, object segmentation, normals, depth, and outlines.
- Easy integration with Unity simulations.
- Customizable settings for camera and image generation.
- Designed to work seamlessly with Python for data loading and machine learning tasks.
- Clone this repository to your local machine.
- Open the project in Unity.
- Set up the cameras and renderers for the desired modalities in your Unity scene.
- Configure the settings for the image generation process.
- Run the simulation to generate the synthetic dataset.
- Integrate the dataset with your Python-based machine learning pipeline.
You can easily customize the dataset by adjusting the camera settings or creating new camera scripts to capture additional modalities. Modify the provided scripts or create new ones to suit your specific needs and requirements.