Multi-modal Instruction Tuned LLMs with Fine-grained Visual Perception
This work addresses the problem of limited visual grounding and interaction flexibility in multimodal AI systems for users needing detailed object-level analysis, representing a novel method for a known bottleneck rather than a foundational breakthrough.
The paper tackles the gap in fine-grained pixel-level perception and multi-modality interaction in multimodal large language models by proposing AnyRef, which generates pixel-wise object perceptions and descriptions from various references like text, boxes, images, or audio, achieving state-of-the-art results on multiple benchmarks.
Multimodal Large Language Model (MLLMs) leverages Large Language Models as a cognitive framework for diverse visual-language tasks. Recent efforts have been made to equip MLLMs with visual perceiving and grounding capabilities. However, there still remains a gap in providing fine-grained pixel-level perceptions and extending interactions beyond text-specific inputs. In this work, we propose {\bf{AnyRef}}, a general MLLM model that can generate pixel-wise object perceptions and natural language descriptions from multi-modality references, such as texts, boxes, images, or audio. This innovation empowers users with greater flexibility to engage with the model beyond textual and regional prompts, without modality-specific designs. Through our proposed refocusing mechanism, the generated grounding output is guided to better focus on the referenced object, implicitly incorporating additional pixel-level supervision. This simple modification utilizes attention scores generated during the inference of LLM, eliminating the need for extra computations while exhibiting performance enhancements in both grounding masks and referring expressions. With only publicly available training data, our model achieves state-of-the-art results across multiple benchmarks, including diverse modality referring segmentation and region-level referring expression generation.