Advancing Fine-Grained Visual Understanding with Multi-Scale Alignment in Multi-Modal Models
This work addresses a specific bottleneck in multi-modal models for fine-grained visual understanding, offering an incremental improvement through enhanced alignment techniques.
The paper tackles the problem of inadequate alignment for fine-grained knowledge in multi-modal large language models, which limits their ability to capture local details and achieve global perception, by introducing a novel fine-grained visual knowledge alignment method that integrates multi-scale object knowledge, resulting in TinyGroundingGPT achieving outstanding results in grounding tasks with performance comparable to larger models.
Multi-modal large language models (MLLMs) have achieved remarkable success in fine-grained visual understanding across a range of tasks. However, they often encounter significant challenges due to inadequate alignment for fine-grained knowledge, which restricts their ability to accurately capture local details and attain a comprehensive global perception. While recent advancements have focused on aligning object expressions with grounding information, they typically lack explicit integration of object images, which contain affluent information beyond mere texts or coordinates. To bridge this gap, we introduce a novel fine-grained visual knowledge alignment method that effectively aligns and integrates multi-scale knowledge of objects, including texts, coordinates, and images. This innovative method is underpinned by our multi-scale fine-grained enhancement data synthesis pipeline, which provides over 300K essential training data to enhance alignment and improve overall performance. Furthermore, we present TinyGroundingGPT, a series of compact models optimized for high-level alignments. With a scale of approximately 3B parameters, TinyGroundingGPT achieves outstanding results in grounding tasks while delivering performance comparable to larger MLLMs in complex visual scenarios.