Skip to content

sushidelivery/yoloface

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

27 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Yolov5 Face Detection

Description

The project is a wrap over yolov5-face repo. Made simple portable interface for model import and inference. Model detects faces on images and returns bounding boxes and coordinates of 5 facial keypoints, which can be used for face alignment.

Installation

pip install -r requirements.txt

Usage example

from face_detector import YoloDetector
import numpy as np
from PIL import Image

model = YoloDetector(target_size=720,gpu=0,min_face=90)
orgimg = np.array(Image.open('test_image.jpg'))
bboxes,points = model.predict(orgimg)

You can also pass several images packed in a list to get multi-image predictions:

bboxes,points = model.predict([image1,image2])

You can align faces, using align class method for predicted keypoints. May be useful in conjunction with facial recognition neural network to increase accuracy:

crops = model.align(orgimg, points[0])

If you want to use model class outside root folder, export it into you PYTHONPATH:

export PYTHONPATH="${PYTHONPATH}:/path/to/yoloface/project/"

or the same from python:

import sys
sys.path.append("/path/to/yoloface/project/")

Other pretrained models

You can use any model from yolov5-face repo. Default models are saved as entire torch module and are bound to the specific classes and the exact directory structure used when the model was saved by authors. To make model portable and run it via my interface you must save it as pytorch state_dict and put new weights in weights/ folder. Example below:

model = torch.load('weights/yolov5m-face.pt', map_location='cpu')['model']
torch.save(model.state_dict(),'path/to/project/weights/yolov5m_state_dict.pt')

Then when creating YoloDetector class object, pass new model name and corresponding yaml config from models/ folder as class arguments. Example below:

model = YoloFace(weights_name='yolov5m_state_dict.pt',config_name='yolov5m.yaml',target_size=720)

Result example

Citiation

Thanks deepcam-cn for pretrained models.

About

Yolov5 Face Detection

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 98.8%
  • Other 1.2%