Simple Deep NN Project

This project is part of an academic project. This Deep NN model has been cloned from github and we are using it to train it on blood cells dataset, instead of the satellite images dataset it was originally developed for. What I need from you is to run the model such that it is able to output the metrics (SSIM, MSE, PSNR) from the new datasets we have. In total there are 8 datasets therefore we will need 16 metrics (3 metrics per dataset). The reason why we have 8 datasets is because we want to proof that the larger the dataset size the more accurate the metrics are. We also introduced different types of noises to the low-resolution images in the dataset used to train the model. What I will provide is the trained model .pth files (for each dataset) and you are required to run the code and output the metrics as well as make graphs of the metrics using tensorboard feature. Please note that the metrics code is already developed and can be found in the evaluator.py and predit.py files.

Using Google Colab