Scaffolding Coordinates to Promote Vision-Language Coordination in Large Multi-Modal Models

TL;DR

This paper introduces Scaffold prompting, a novel visual prompting scheme that overlays a dot matrix and uses coordinate references to enhance vision-language coordination in large multi-modal models, especially in complex reasoning tasks.

Contribution

The paper proposes Scaffold prompting, a simple and general visual prompting method that improves vision-language coordination in large multi-modal models by overlaying visual anchors and using coordinate references.

Findings

Scaffold outperforms GPT-4V with textual CoT prompting on various tasks.

The method enhances reasoning capabilities in challenging vision-language scenarios.

Extensive experiments validate the effectiveness of Scaffold prompting.

Abstract

State-of-the-art Large Multi-Modal Models (LMMs) have demonstrated exceptional capabilities in vision-language tasks. Despite their advanced functionalities, the performances of LMMs are still limited in challenging scenarios that require complex reasoning with multiple levels of visual information. Existing prompting techniques for LMMs focus on either improving textual reasoning or leveraging tools for image preprocessing, lacking a simple and general visual prompting scheme to promote vision-language coordination in LMMs. In this work, we propose Scaffold prompting that scaffolds coordinates to promote vision-language coordination. Specifically, Scaffold overlays a dot matrix within the image as visual information anchors and leverages multi-dimensional coordinates as textual positional references. Extensive experiments on a wide range of challenging vision-language tasks demonstrate the superiority of Scaffold over GPT-4V with the textual CoT prompting. Our code is released in https://github.com/leixy20/Scaffold.