Skip to content

Latest commit

 

History

History
7 lines (6 loc) · 437 Bytes

File metadata and controls

7 lines (6 loc) · 437 Bytes

WhoTube -> Predict Youtubers based on Youtube Thumbnails

Web based Flask Aplication deployed -> https://www.angrybruce.com/

  1. Youtube Thumbnails Extractor : Download Thumbnail based on YT channel IDs
  2. CNN Training for Colab : Train Xception with ImageNet weights + Image Generators. Apply GradCam to understand the CNN "focus".
  3. Flask Webapp to serve the Model
  4. Out Scope : Deployment in VM with nginx reverse proxy + gunicorn