You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Solutions for Advanced Image Analysis course assignments, featuring model designs for image summation and generation with MNIST, and style transfer using CycleGAN with MNIST and SVHN datasets.
This project implements a CycleGAN-based method for image dehazing on a paired dataset containing hazy images and their corresponding ground truth clear images.
We train a CycleGAN model that will generate "realistic" augmented images based on images coming from the Duckietown simulator. This is in an attempt to reducing the reality gap when transitioning from robot training in simulation to real life.