Up Scaling Video

Phase 1: Dataset Creation

• Downloading videos from YouTube in 1080p and 480p
• Splitting videos in frame with ffmpeg, one frame per second
• Copy frames in a good format for training

Phase 2: Model Training (v2)

• Buffer images with the Tensorflow model
• Create a model which fit in AWS instances
• Train the model
• Test the model
  • Split a full video in frame
  • Predict the high resolution of all images
  • Merge predicted images