Retrieval-Based Interleaved Visual Chain-of-Thought in Real-World Driving Scenarios
This addresses the challenge of enhancing VLM reasoning in complex real-world scenarios like driving, though it is incremental as it builds on existing CoT methods.
The paper tackled the problem of limited visual reasoning in vision-language models (VLMs) by introducing RIV-CoT, a retrieval-based interleaved visual chain-of-thought method, which improved answer accuracy by 3.1% and reasoning accuracy by 4.6% over vanilla CoT prompting on a driving dataset.
While chain-of-thought (CoT) prompting improves reasoning in large language models, its effectiveness in vision-language models (VLMs) remains limited due to over-reliance on textual cues and memorized knowledge. To investigate the visual reasoning capabilities of VLMs in complex real-world scenarios, we introduce DrivingVQA, a visual question answering dataset derived from driving theory exams, which contains 3,931 multiple-choice problems with expert-written explanations and grounded entities relevant to the reasoning process. Leveraging this dataset, we propose RIV-CoT, a Retrieval-Based Interleaved Visual Chain-of-Thought method that enables VLMs to reason using visual crops corresponding to these relevant entities. Our experiments demonstrate that RIV-CoT improves answer accuracy by 3.1% and reasoning accuracy by 4.6% over vanilla CoT prompting. Furthermore, we demonstrate that our method effectively scales to the larger A-OKVQA reasoning dataset by leveraging automatically generated pseudo-labels, outperforming CoT prompting.