We present Set-of-Mark (SoM), a new visual prompting method, to unleash the visual grounding abilities of large multimodal models (LMMs), such as GPT-4V. As illustrated in Fig. 1 (right), we employ off-the-shelf interactive segmentation models, such as SEEM/SAM, to partition an image into regions at different levels of granularity, and overlay these regions with a set of marks e.g., alphanumerics, masks, boxes. Using the marked image as input, GPT-4V can answer the questions that require visual grounding. We perform a comprehensive empirical study to validate the effectiveness of SoM on a wide range of fine-grained vision and multimodal tasks. For example, our experiments show that GPT-4V with SoM in zero-shot setting outperforms the state-of-the-art fully-finetuned referring expression comprehension and segmentation model on RefCOCOg. Code for SoM prompting is made public at: https://github.com/microsoft/SoM.

翻译：我们提出了一种新的视觉提示方法——Set-of-Mark（SoM），用于激发大型多模态模型（LMMs，如GPT-4V）的视觉定位能力。如图1（右）所示，我们利用现成交互式分割模型（如SEEM/SAM）将图像按不同粒度层级分割成若干区域，并为这些区域叠加一组标记（例如字母数字符号、掩码、边界框）。通过将带标记的图像作为输入，GPT-4V能够回答需要视觉定位的问题。我们进行了全面的实证研究，验证了SoM在广泛细粒度视觉与多模态任务中的有效性。例如，实验表明，在零样本设置下，使用SoM的GPT-4V在RefCOCOg基准上超越了现有最先进的经过全微调的指代表达理解与分割模型。SoM提示的代码已开源：https://github.com/microsoft/SoM。