In recent years, deep learning has attracted great attention, thanks to its impressive results in a wide range of tasks. The first and main application of modern deep learning was object recognition, but all aspects of image/video processing and computer vision are strongly involved and have witnessed great progress in performance. This course will be organized in two main parts
1) A review of the fundamentals of deep learning, history, main types of deep networks, popular architectures, and main approaches for training them;
2) A focus on successful solutions to fundamental image processing problems, such as image denoising, de-blurring, super-resolution, segmentation, classification and fusion, generation, integrity verification (multimedia forensics).
The course will include some laboratory work in Python and TensorFlow.