GitHub / momothain / cs143-final-facebox-anime-stylize

Our project sought to combine two things: face recogni- tion and style transfer to carry over anime character quali- ties onto those recognized faces. We used a pre-trained DNN from OpenCV to recognize and locate the faces, cropped those, used the cropped image as the content for style trans- fer using a pre-trained VGG, and overlaid that output on the original image. This VGG came from TensorFlow’s models, specifically a VGG-19 trained on ImageNet.

JSON API: https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/momothain%2Fcs143-final-facebox-anime-stylize

Files

Dependencies