New Dataset and Methods for Fine-Grained Compositional Referring Expression Comprehension via Specialist-MLLM Collaboration
This work addresses the challenge of fine-grained reasoning in vision-language tasks for researchers and practitioners, offering incremental advancements through dataset enhancements and model collaboration.
The authors tackled the problem of fine-grained compositional referring expression comprehension by introducing a new dataset with controllable difficulty and negative examples, and proposed two collaborative methods between specialist models and MLLMs, resulting in significant improvements on their dataset and other benchmarks.
Referring Expression Comprehension (REC) is a foundational cross-modal task that evaluates the interplay of language understanding, image comprehension, and language-to-image grounding. It serves as an essential testing ground for Multimodal Large Language Models (MLLMs). To advance this field, we introduced a new REC dataset in our previous conference paper, characterized by two key features. First, it is designed with controllable difficulty levels, requiring multi-level fine-grained reasoning across object categories, attributes, and multi-hop relationships. Second, it incorporates negative text and images generated through fine-grained editing and augmentation, explicitly testing a model's ability to reject scenarios where the target object is absent, an often overlooked yet critical challenge in existing datasets. In this extended work, we propose two new methods to tackle the challenges of fine-grained REC by combining the strengths of Specialist Models and MLLMs. The first method adaptively assigns simple cases to faster, lightweight models and reserves complex ones for powerful MLLMs, balancing accuracy and efficiency. The second method lets a specialist generate a set of possible object regions, and the MLLM selects the most plausible one using its reasoning ability. These collaborative strategies lead to significant improvements on our dataset and other challenging benchmarks. Our results show that combining specialized and general-purpose models offers a practical path toward solving complex real-world vision-language tasks. Our dataset and code are available at https://github.com/sleepyshep/FineCops-Ref.