OmniCap-IF: Benchmarking and Improving Instruction Following Abilities for Omni-Video Captioning

jiahaowang, An Ping, Yanghai Wang, Yuanxing Zhang, Shihao Li, Hanyan Bian, Yichi Ren, Yize Zhang, Han Wang, Haowen Chen, Junze Li, Jiaqi Wang, Yiyang Hu, Zhuze Xu, Zijie Zhang, Jiaheng Liu

OmniCap-IF: Benchmarking and Improving Instruction Following Abilities for Omni-Video Captioning: 13 upvotes on Hugging Face Daily Papers, #18 of 59 papers on 2026-06-09. Day-by-day upvote history.

While Omni-modal Large Language Models (OLLMs) have demonstrated impressive capabilities in jointly processing audio and visual streams, their ability to strictly adhere to complex, multi-faceted user instructions remains largely unexplored. Existing benchmarks primarily focus on holistic video understanding or text-only instruction following, failing to capture the intricate interplay between modalities and user constraints. To bridge this gap, we introduce OmniCap-IF, the first comprehensive benchmark specifically designed to evaluate instruction-following capabilities in omni-modal captioning. OmniCap-IF incorporates a systematic framework that assesses captions on two dimensions: format correctness and content correctness. Our benchmark encompasses 50 distinct constraint types across pure visual, pure audio, and audio-visual modalities, while integrating Temporal Grounding to assess spatio-temporal precision. Extensive evaluations of prominent models on 1,920 high-quality samples reveal significant performance disparities. Furthermore, our analysis uncovers a critical "format-content tradeoff", demonstrating that increasing formatting complexity directly degrades models' omni-modal reasoning abilities. Finally, to advance the field, we curate a 54K instruction-tuning dataset, OmniCap-IF-54K and present OmniCaptioner-IF, which achieves notable improvements in both complex instruction adherence and general omni-modal captioning performance.

Paper page on Hugging Face · arXiv

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.