The original work of this project is always belong to the original creator (https://github.com/TropComplique/mtcnn-pytorch). I just make this available on pypi for easy installation. To install this project just type pip install torch-mtcnn

MTCNN

pytorch implementation of inference stage of face detection algorithm described in
Joint Face Detection and Alignment using Multi-task Cascaded Convolutional Networks.

Example

How to use it

Install the package with pip: pip install torch-mtcnn

from torch_mtcnn import detect_faces
from PIL import Image

image = Image.open('image.jpg')
bounding_boxes, landmarks = detect_faces(image)

For a few more examples available on the original repository (link above).

Requirements

• pytorch 1.1.0^
• Pillow, numpy

Credit

This implementation is heavily inspired by:

• pangyupo/mxnet_mtcnn_face_detection