How to use Before / After images to train a model


I am trying to create a model that can clean pictures of noise, blur, high luminosity etc, but I do not know how to do that. I have tried to search for it a lot, and I couldn't find anything that could properly teach me, as I am very new in this.

I have tried to do something using Google Colaboratory so that people can look it up and help me a bit: Image Cleaning Modedl

At first, I download a file containing 5000+ clean pictures and their degraded version. Then, I load a training set of 500 pictures in two ndarray datasets called degraded_dataset and clean_dataset of shape=(500, 576, 720, 3).

Then, I do not know what to do with these two datasets, I do not know how to train a model. I vaguely have an idea, which is giving the clean_dataset to the model as the label_dataset, however I am not sure it is the good way of doing it.

I followed these tutorials: Tensorflow CNN & Kaggle "From images to narray"

Thank you for helping me.


Posted 2020-01-20T01:33:36.197

Reputation: 31

Hi I think you might try asking on StackOverflow. This community doesn't focus on implementation (see I'd also recommend A. Ng Deep Learning Specialization to learn how to do these sort of things.

– respectful – 2020-01-20T03:26:36.043

No answers