Visual Training of a CNN as a Denoiser (Deep Image Prior)

preview_player
Показать описание
Visualisation of training epochs for a convolutional neural network (CNN) serving as a deep image prior (DIP), which can find application in image denoising.

No training data is needed, only the image data to process. Hence this is a very simple AI model for image processing, great as an introduction to deep learning and artificial intelligence (AI).

The deep image prior was proposed by Ulyanov, Vedaldi and Lempitsky in 2017.
Рекомендации по теме