llm-lab-org/Multimodal-RAG-Survey

Organizes and taxonomizes papers on multimodal RAG systems across retrieval strategies (text/vision/video/audio-centric), fusion mechanisms, augmentation techniques, and generation approaches. Provides comprehensive dataset benchmarks spanning image-text, video, audio, medical, and fashion domains with evaluation metrics and training methodologies. Continuously updated resource tracking advances in cross-modal alignment, agentic interaction, and robustness for systems that ground LLM outputs in multimodal external knowledge bases.

487 stars.