tencent-ailab / IP-Adapter

The image prompt adapter is designed to enable a pretrained text-to-image diffusion model to generate images with image prompt.
Apache License 2.0
5.08k stars 331 forks source link

insightface-faceid #275

Open JackAILab opened 8 months ago

JackAILab commented 8 months ago

When I used the insightface module to extract faceids from the ffhq and celeba data sets, I found that face IDs could be extracted from 70,000 faces in ffhq, but face IDs could not be extracted (faces result is None: []) from more than 1,000 faces in the celeba data set. What should I do? And when I used IPA-FacePlus for actual inference, some uploaded faces could not extract faceid. It seemed that some changes could be made, but there was no unified method suitable for all images. If you can provide some ideas, thank you very much!