Extract Image frames from video to apply neural style and save converted images as gif
- Trained a CNN model ( VGG19 architecture before flattening ) on orignal image and style image as per paper on neural style transfer .
- Extracted image frames from video .
- Calulated content loss between original image and at every layer image pixel value.
- Calculated gram matrix over style image and find style loss between gram matrix and at every layer image pixel value.
- After every iteration style loss decreases as style is being imposed on original image.
- All converted images were saved as gif.
Video used for extracting image frames and applying different style
Video.mp4
GIF formed from converted images
