Visual Chain-of-Thought
An extension of textual chain-of-thought to the visual modality, in which a model generates a series of intermediate visual artifacts, such as local annotations, crops and sketches, as thinking drafts before producing a final answer, thereby supporting spatial and physical reasoning.
Timeline (last 90 days)
Visual Chain-of-Thought (Visual CoT) has the model generate intermediate reasoning images while producing intermediate text steps, such as annotating key regions, predicting future frames, or drawing spatial relationship diagrams.
If Visual CoT needs to generate 3–5 intermediate images in a single inference pass, the cumulative latency can reach 10–40 seconds.
视觉推理链（Visual Chain-of-Thought）借鉴语言模型的思维链提示技术，将复杂视觉问题分解为多个步骤逐步求解
All Facts (3)
Visual Chain-of-Thought (Visual CoT) has the model generate intermediate reasoning images while producing intermediate text steps, such as annotating key regions, predicting future frames, or drawing spatial relationship diagrams.
50%UnverifiedIf Visual CoT needs to generate 3–5 intermediate images in a single inference pass, the cumulative latency can reach 10–40 seconds.
50%Unverified视觉推理链（Visual Chain-of-Thought）借鉴语言模型的思维链提示技术，将复杂视觉问题分解为多个步骤逐步求解
50%