This repository provides a comprehensive survey of 3D Gaussian Splatting (3DGS) applications, focusing on segmentation, editing, and generation. It aims to systematically review downstream applications beyond classical view synthesis, serving as a valuable resource for researchers and practitioners in computer vision and graphics seeking to understand the state-of-the-art in these rapidly evolving areas.

How It Works

The survey systematically categorizes and reviews existing research within three primary application domains: segmentation, editing, and generation of 3D Gaussian Splatting. It details relevant datasets, summarizes key methodologies, and provides links to associated projects or code where available, offering a structured overview of advancements and trends in extending 3DGS capabilities.

Quick Start & Requirements

This repository hosts a survey paper and does not contain executable code or software requiring installation. It serves as a curated collection of research findings and resources.

Highlighted Details

• Extensive cataloging of datasets used for 3DGS segmentation, editing, and generation, including details on venue, scene/view counts, and highlights.
• Comprehensive lists of methods for each application area, detailing publication year, venue, paper abbreviation, title, and code availability.
• Coverage extends to other application tasks such as Human Avatar reconstruction, SLAM, and 3DGS Detection.
• The survey includes numerous citations and links to related survey papers for further reading.

Maintenance & Community

The repository welcomes community contributions through issues and pull requests to include any missed relevant works, indicating an effort to maintain the survey's completeness over time.

Licensing & Compatibility

No specific software license is mentioned for the repository's content. The survey itself is presented as a research publication.

Limitations & Caveats

The survey's content is based on research available up to its publication date (indicated as 2025 in the citation). Not all methods listed have publicly available code, as denoted by '❌' in the project/code columns. The primary focus is on segmentation, editing, and generation, with other applications covered more briefly.