Computational Imaging Lab Activities

Lecturers:   Luca Ratti  and  Juliàn Tachella

This laboratory complements the topics of the other teaching units by providing tools and examples to try implementing some learning-based techniques for image processing. The first session is designed as a numerical counterpart to the pre-course, focusing on basic image processing in Python and on image denoising, showcasing how to define and train a neural network denoiser on a handcrafted supervised dataset. The second session is a hands-on tutorial on Deep Inverse, a Pytorch library that provides a seamless integration of deep learning techniques with imaging inverse problems. The session will show how to consider several ill-posed forward operators and introduce variational and network-based reconstruction methods.