Instructions to use CrucibleAI/ControlNetMediaPipeFace with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use CrucibleAI/ControlNetMediaPipeFace with Diffusers:
pip install -U diffusers transformers accelerate
from diffusers import ControlNetModel, StableDiffusionControlNetPipeline controlnet = ControlNetModel.from_pretrained("CrucibleAI/ControlNetMediaPipeFace") pipe = StableDiffusionControlNetPipeline.from_pretrained( "stabilityai/stable-diffusion-2-1-base", controlnet=controlnet ) - Notebooks
- Google Colab
- Kaggle
In the newest controlnet1.1,the problem of identify exception exists;
#17
by CGLboulder - opened
Sure. The original batch of images was created on our branch of the ControlNet repo: https://github.com/crucible-ai/ControlNet/ with the gradio_face2image.py script. If MediaPipe is missing faces, it's likely because of cropping or resizing issues. The face detection component is particularly sensitive to "canvas size" settings. If a face is not detected and the aspect ratio is non-square, it can be worth manually padding the image out to square. Are you using the automatic1111 webui?

