ASID-Caption

ASID-1M: a large-scale collection of attribute-structured audiovisual instructions with both single-attribute and all-attributes training formats.
ASID-Verify: a scalable curation pipeline that generates, ensembles, verifies, and refines annotations to improve semantic and temporal consistency.
ASID-Captioner: Qwen2.5-Omni-based audiovisual captioning models fine-tuned on ASID-1M.

community

Video Understanding, Audio-Visual Learning, Multimodal LLMs, Video Captioning, Instruction Tuning, Dataset Curation

lyhisme submitted a paper 1 day ago

lyhisme updated a dataset 1 day ago

lyhisme updated a model 1 day ago

Towards Universal Video MLLMs with Attribute-Structured and Quality-Verified Instructions

Submitted by

Yunheng Li