Thesis Proposals

GPU programming with CUDA and optimization of Image Processing algorithms with particular attention to Connected Components Labeling and Thinning.

1. Programming Guide

1. Optimized Block-Based Algorithms to Label Connected Components on GPUs, IEEE Transactions on Parallel and Distributed Systems
2. A New Algorithm for Parallel Connected-Component Labelling on GPUs, IEEE Transactions on Parallel and Distributed Systems
3. How does Connected Components Labeling with Decision Trees perform on GPUs?, 18th International Conference on Computer Analysis of Images and Patterns

This thesis aims to develop a django-based web application to collect, preview, annotate, and distribute medical datasets. The application design can be inspired by the ISIC archive website and must support 2D images, Virtual Slides, and 3D volumes.

1. Tutorial

Analysis and development of Deep Learning techniques for medical imaging. The activity can be carried out on different research topics:

1. Segmentation and classification of skin lesions from dermoscopic and/or confocal images



2. Inferior alveolar nerve canal detection from dental Cone Beam Computed Tomography (CBCT)



3. Glomeruli identification in Virtual Slide images

1. Paper
2. ISIC Dataset
3. Challenge 2020

1. Paper 1
2. Paper 2

1. Paper 1
2. Dataset
3. Paper 2

Unit Testing is a software test level in which individual units/components of a software are tested. The aim is to validate that each unit of the software functions as expected. This thesis aims at developing unit test functionalities for the Computer Vision library ECVL, by using the googletest framework. The ECVL library is under development by UNIMORE within the H2020 European Project DeepHealth.

1. Source Code
2. Documentation

1. Source Code
2. Tutorial
3. Examples

1. The DeepHealth Toolkit: A Unified Framework to Boost Biomedical Applications.

Missioni (missioni.ing.unimore.it) is a Django-based web application for the semi-automatic travel reimbursement forms filling. This thesis aims at developing a mobile version of such a service, possibily integrating existing systems for the acquisition of data directly from receipt photos.

1. Source Code
2. Website

1. Tutorial

This thesis aims to develop a python script for the web scraping of exam dates (using for example selenium web driver) and their publication on a dedicated google calendar.

1. Tutorial

1. Tutorial

Development of a Django-based web application to perform inference on existing and previously trained Deep Learning models. The final goal is to define/implement a mechanism to connect the application (i.e. the individual models/architectures available) to GitHub repositories, so that each update on a specific repository will be automatically reflected and available within the web application.

1. Tutorial

1. Tutorial

1. Tutorial