github-mirrors/face-study

mirror of https://github.com/vipul-sharma20/face-study.git synced 2026-06-28 08:33:10 +00:00

Study of face data set

Python 100%

Find a file

Vipul Sharma c6f1aaf131 added example		2016-06-30 00:05:10 +05:30
detect.py	add rectangle and contour based visualization	2016-06-27 12:36:30 +05:30
haarcascade_frontalface_alt.xml	code for mouth detection added	2016-03-16 18:39:58 +05:30
mouth_cascade.xml	code for mouth detection added	2016-03-16 18:39:58 +05:30
README.md	added example	2016-06-30 00:05:10 +05:30

README.md

Face-study

~~Study of neutral face dataset from: http://agingmind.utdallas.edu/facedb/view/neutral-faces~~ [URL DEAD]
~~Dataset file: Color Neutral jpg (11MB)~~ [URL DEAD]
Processed dataset: Color Neutral Processed (23MB)
Accuracy: 82%
Face Area : Mouth Area (Ratio) (bounding rectangle method) Average: 22.39
Face Area : Mouth Area (Ratio) (lip contour method) Average: 54.30

Sample dataset example

Sample dataset

Detect face and mouth using Haar Feature-based Cascade Classifier
- Detect face
- Extract face and detect mouth
- Correct detection
Extracted mouth region

Thresholding to extract lips (Otsu's Binarization Method)

Shi-Tomasi Corner Detection (In progress)

Complete processed image (contours + Shi-Tomasi corner detection)