This project is depreciated and has been replaced by SemanticSegmentation!!!!
Or ObjectDetection if you want to find instances of skin!
This is a high-speed python based skin detection system using OpenCV, it is done using adaptive thresholding, reference papers can be found below. It is designed for processing VGA sized images in real time for Gesture Control.
Getting the app to run is pretty easy. This script will not install OpenCV. However to install the rest of the project dependencies and run the demo script use the following commands;
# Clone the repo
git clone https://github.com/WillBrennan/SkinDetector && cd SkinDetector
make install
python FromFile.py <directory of images> --displayimport cv2
import skin_detector
img_path = raw_input("Please Enter Image Path")
image = cv2.imread(img_path)
mask = skin_detector.process(image)
cv2.imshow("input", image)
cv2.imshow("mask", mask)
cv2.waitKey(0)
This repo is based upon the paper;
- Skin Segmentation Using Multiple Thresholds (2003) - http://www.ivl.disco.unimib.it/papers2003/EI06-EI109%20Skin-paper.pdf
. Which is no longer accessiable; its similar to the technique used in the more modern paper, Skin color segmentation using multi-color space threshold.