Unverified50% confidenceFactExact time
The portion estimation accuracy of photo-based nutrition recognition typically has a margin of error of 20–30%.
1
Sources
50%
Confidence
Long-term
Relevance
8/10/2026
First Seen
Sources
Macrobite：拍照即算营养，让宏量素追踪不再繁琐
Related Claims
UnverifiedEarly CNN-based food image classification solutions achieved accuracy of only approximately 70–80%.64% similarUnverified推荐为每个人注册5-10张不同条件下的照片，使用多模板平均向量可将识别准确率提升2-5个百分点63% similarUnverified业界仿真精度评估通常要求吞吐量误差小于15%63% similarVerifiedIf the accuracy per step is 95%, after 20 steps the overall accuracy will drop to approximately 36%.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/726425API
curl https://kongchang.com/api/v1/knowledge/claims/726425MCP
get_claim(id=726425)