Toward Low-Burden Image-Based Dietary Assessment Through Human-AI Collaborative Correction
Human-AI Collaborative Correction is introduced, which reframes correction as communicate intent regenerate $\rightarrow$ regenerate $\rightarrow$ verify: users express what should change through text, voice, or guided clarification, the AI regenerates the linked dietary record, and users verify the result.
Darren Liu, Andrew Lu, ZhiyuanRyanWang et al.
· 0 citations