FLARE

FLARE: Full-Modality Long-Video Audiovisual Retrieval Benchmark with User-Simulated Queries 🤗 About This Benchmark This repository hosts the full release of FLARE: Full-Modality Long-Video Audiovisual Retrieval Benchmark with User-Simulated Queries. FLARE screens 399 long-form videos (10–60 min, 225.4 h total) from Video-MME and segments them into 87,697 fine-grained clips, each annotated with three captions — vision-only, audio-only, and unified audiovisual — and… See the full description on the dataset page:

Source: Hugging Face Hub (AnonymousFLARE/FLARE). Metadata imported from the dataset’s Hub tags.