2026.06.03
Our paper “SetPieceRAG: Domain-Specific RAG for Knowledge-Intensive Soccer VQA with Large Language Models” was accepted to the CVPR 2026 Workshop on Computer Vision in Sports (CVSports) as a spotlight presentation, and Young Seon Kim presented the work in Denver, Colorado.
SetPieceRAG is a domain-specific Retrieval-Augmented Generation (RAG) framework tailored for soccer VQA. Beyond the core RAG mechanism, it introduces task-specific adaptations including LLM ensembling, LoRA-based domain fine-tuning, super-resolution preprocessing, and object-centric analysis via SAHI.
To tackle knowledge-intensive queries, the approach integrates vision-language tools like CLIP with external knowledge sources — a domain-specific corpus (SoccerWiki) and the web-search capabilities of multimodal LLMs — providing the non-parametric and long-tail knowledge that language models inherently lack. The system sets state-of-the-art results on the challenging SoccerNet VQA benchmark.