这对吗？用的纳米 app 妆容迁移智能体

[AIFSH]

妆容

素图

结果

[liushanyuan18]

Thanks for the feedback! This is a very valuable observation.

Currently, the model does have limitations when handling "hair-on-face" occlusions. This is mainly because the training dataset contained a relatively low proportion of samples with hair overlapping the face.

We plan to optimize and upgrade the model to address this in future iterations.