“Set-Of-Mark Prompting Unleashes Extraordinary Visual Grounding in GPT-4-V”, Jianwei Yang, Hao Zhang, Feng Li, Xueyan Zou, Chunyuan Li, Jianfeng Gao2023-10-17 (GPT-4 nonfiction)⁠:

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-4-V.

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 eg. alphanumerics, masks, boxes. Using the marked image as input, GPT-4-V 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-4-V 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: Github.