/Facial-Keypoints-Detection

This project uses LeNet inspired CNN architecture to detect 68 facial keypoints in an image. Facial keypoints include points around eyes, noses, eyebrows, mouth and facial border. Facial Keypoints have applications in face morphing, emotion recognition, face-tracking, face-pose recognition, facial filters and many more...

Primary LanguageJupyter Notebook

Facial Keypoints Detection



Description

This project uses LeNet inspired CNN architecture to detect 68 facial keypoints in an image. Facial keypoints include points around eyes, noses, eyebrows, mouth and facial border. Facial Keypoints have applications in face morphing, emotion recognition, face-tracking, face-pose recognition, facial filters and many more...

Requirements

  • Tensorflow
  • OpenCV
  • skimage

Dataset

Youtube faces dataset is used for training/testing and can be accesed from here. There are 3,462 training images and around 2000 test images. Augmentations used : Random Crop and 5 degrees clockwise, anticlockwise rotation

Results

screen shot 2017-08-07 at 12 18 15 pm screen shot 2017-08-07 at 12 18 15 pm
screen shot 2017-08-07 at 12 18 15 pm screen shot 2017-08-07 at 12 18 15 pm
screen shot 2017-08-07 at 12 18 15 pm screen shot 2017-08-07 at 12 18 15 pm
screen shot 2017-08-07 at 12 18 15 pm screen shot 2017-08-07 at 12 18 15 pm
screen shot 2017-08-07 at 12 18 15 pm screen shot 2017-08-07 at 12 18 15 pm
Readme and repo will be updated with more details soon ....