/gaze_estimation

Primary LanguageJupyter Notebook

gaze_estimation

Computer Vision technique helps to trace the position of pupil as well as perform blink detection on pre-recorded videos.

Features

  • Gaze Estimation System can be used in tracking pupil and estimating the number of blinks.
  • This is beneficial for people who cannot speak or use their hands to communicate.
  • It can be useful in identifying diseases such as Parkinson's Disease etc.

Under the hood

  • OpenCV library is used for capturing the video through webcam and reading the frames as well as finding contours on it to analyze the area occupied by pupil.
  • Gaussian blur is used to blur the frames to improve detections and we tried to binarize the frames.
  • We also tried to calculate the aspect ratio of eyes through opencv functions.

OUTPUT VIDEO FRAME SAMPLE