||Gustav Forslid, Håkan Wieslander, David Ramnerö, Nils Bäckström (CBA)
||Master thesis presentation
||Master's thesis project presentations: CNN:s in computer vision
||This seminar will consist of 3 different computer vision master thesis projects utilizing Convolutional Neural Networks.
David is creating a two network pipeline to automatize parts of the training data generation process, in order to decrease amount of manual annotation work required.
Nils has investigated how transfer learning, arithmetic precision, and parameter pruning affects classification results for CNN:s targeted at deployable systems.
Gustav and Håkan have been working on classifying cells from the uterine cervix and oral cavity. The goal is to evaluate the performance of different CNN:s as a part of screening programs for cervical and oral cancer.