This repository serves as a curated list of research papers, datasets, and applications related to Scene Graph Generation (SGG). It targets researchers and practitioners in computer vision and natural language processing, providing a comprehensive overview of the field and its advancements.

How It Works

The repository categorizes SGG research across various modalities (2D, 3D, video, audio, text) and subtasks (predicate classification, scene graph classification, scene graph detection). It also highlights applications of SGG, such as image retrieval, captioning, visual reasoning, and 3D scene generation, offering a structured view of the evolving landscape.

Quick Start & Requirements

This repository is a curated list and does not require installation or execution. It provides links to external resources for datasets and research papers.

Highlighted Details

• Extensive categorization of SGG research by modality, task, and application.
• Detailed tables comparing various scene graph datasets, including their modalities, object/relation classes, and instance counts.
• Inclusion of recent advancements, such as LLM-based approaches and novel datasets like STAR and ReCon1M for satellite imagery.
• Coverage of diverse applications, from image generation and visual reasoning to privacy-sensitive object identification and dynamic environment guidance.

Maintenance & Community

The repository is maintained by ChocoWu. It includes links to relevant workshops and surveys, indicating community engagement and ongoing research efforts.

Licensing & Compatibility

The repository itself is not software and thus does not have a software license. Links provided point to various external resources, each with its own licensing.

Limitations & Caveats

As a curated list, the repository's content is dependent on the availability and accuracy of the linked external resources. It does not provide code for any of the mentioned papers or datasets.