Unverified50% confidenceFactExact time
Set-of-Mark (SoM) is a visual prompting method proposed by Microsoft Research that overlays numeric labels or colored bounding boxes on screenshots to annotate interactive elements, transforming the visual localization problem into a selection problem.
1
Sources
50%
Confidence
Long-term
Relevance
8/10/2026
First Seen
Sources
AI浏览器与电脑自动化模型对比：Claude、Operator及开源方案选型指南
Related Claims
UnverifiedAnnotate App的视觉标注模式与计算机视觉领域的Set-of-Mark（SoM）提示技术有相似之处71% similarUnverifiedThe Zhang's calibration method was proposed by Dr. Zhengyou Zhang of Microsoft Research in 1999, requiring only a flat checkerboard photographed from different angles in several shots to complete calibration.60% similarUnverified图像条件化方案通常借助CLIP视觉编码器（ViT架构）将参考图映射到与文本嵌入共享的语义空间，再通过交叉注意力（Cross-Attention）机制将视觉特征注入UNet的去噪过程58% similarUnverified工作流中的抠图节点依赖语义分割模型（如SAM、U²-Net等），能够自动识别图像中的主体对象并去除背景，输出带透明通道的PNG图像56% similarUnverified参考图嵌入技术栈使用CLIP图像编码器或ArcFace模型将角色参考图编码为高维特征向量，再通过适配层映射到扩散模型条件空间以交叉注意力或特征注入方式施加约束56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/725444API
curl https://kongchang.com/api/v1/knowledge/claims/725444MCP
get_claim(id=725444)