Omni-RGPT: Unifying Image and Video Region-level Understanding via Token Marks
This work addresses the challenge of robust region-level comprehension in multimodal AI systems for applications like video analysis and image understanding, representing an incremental advancement through novel integration techniques.
The authors tackled the problem of inconsistent region-level understanding across images and videos by developing Omni-RGPT, a multimodal large language model that uses Token Marks to unify visual and text representations, achieving state-of-the-art results on commonsense reasoning benchmarks and strong performance in captioning and referring expression comprehension tasks.
We present Omni-RGPT, a multimodal large language model designed to facilitate region-level comprehension for both images and videos. To achieve consistent region representation across spatio-temporal dimensions, we introduce Token Mark, a set of tokens highlighting the target regions within the visual feature space. These tokens are directly embedded into spatial regions using region prompts (e.g., boxes or masks) and simultaneously incorporated into the text prompt to specify the target, establishing a direct connection between visual and text tokens. To further support robust video understanding without requiring tracklets, we introduce an auxiliary task that guides Token Mark by leveraging the consistency of the tokens, enabling stable region interpretation across the video. Additionally, we introduce a large-scale region-level video instruction dataset (RegVID-300k). Omni-RGPT achieves state-of-the-art results on image and video-based commonsense reasoning benchmarks while showing strong performance in captioning and referring expression comprehension tasks.