demanr/smile-detector

Smile Detector

This Python program uses cv2 to detect smiles from images and webcam.

It converts face data into greyscale for improved classification, and was made with reference to a Youtube livestream by Clever Programmer, which can be found here https://www.youtube.com/watch?v=uLY5JSE5WAU.

Dataset and Haar Cascades

The generated_haarcascade_grey_smile dataset was created using images from the GENKI4K dataset, which were then converted to greyscale for improved performance. This dataset was generated using Cascade Trainer GUI.

The haarcascade_frontalface_default and haarcascade_smile xml files are two of the OpenCV Haar Cascades, which can be found on the OpenCV Github under data.

Program Use

If the generated AI is having a hard time reading the face, ensure the face is in a bright area, and no parts are blocked from view.

To display the detected image(s) the program will open a tab. To close these tabs, press ESCAPE and alternatively for photo analysis click the X.

Image

Example image by leah hetteberg on Unsplash.