Referring Multiple Regions with Large Multimodal Models via Contextual Latent Steering
For practitioners needing precise region-level visual understanding without costly fine-tuning, CSteer offers a training-free solution that surpasses specialized models.
The paper introduces Contextual Latent Steering (CSteer), a training-free method that enables large multimodal models to refer to multiple regions contextually, outperforming tailored referring models on most benchmarks and setting new state-of-the-art results.
Large Multimodal Models (LMMs) have recently demonstrated their proficiency in holistic visual comprehension. However, most of them struggle to tackle region-level perception guided by visual prompts, especially for cases where multiple regions are referred simultaneously, or scenarios where global contexts are necessary for precise visual referring. We introduce Contextual Latent Steering (CSteer), a training-free approach for guiding general LMMs to refer multiple regions contextually, without expensive fine-tuning or architectural modifications. CSteer starts with pre-computing contextual vectors that implicitly represent visual referring behaviors, such as differentiation among regions and attention to global contexts, followed by representation editing during inference time. Experimental results on multiple datasets indicate that general LMMs with CSteer outperform tailored referring LMMs in most cases, suggesting a promising solution in training-free, and setting new state-of-the-art for this field. Code is available at https://github.com/xing0047/csteer.git.