Prof. Marcus Hudritsch

Studiengang für Informatik an der BFH

CAS Practical Machine Learning | PML

BTI4402 Introduction to Image Processing and Artificial Intelligence

BTI4404 Advanced Image Processing and Artificial Intelligence

BTI4406 Special Topics in Image Processing and Artificial Intelligence

BTI7527 Game Development

BTI7534 Advanced Game Development

Applied Research and Development Projects

Projects focusing Augmented Reality

Projects focusing Virtual Reality

Development Project SLProject on Github

Development Project ImagePlay on Github

Image Analysis and Image Synthesis

2D & 3D Computer Graphics

Augmented and Virtual Reality

Image based Machine Learning

2021-2022: SmartBreed: Using automated breeding boxes, farmers can breed grasshoppers on their farms. These convert agricultural by-products into chickens' protein, improving their health and contributing to resilient circular regional production. The breeding boxes have been clearly improved in terms of material, breeding volume and functionality. The manufacturing costs were mainly due to: Massive raw material price increases did not reduce below 750 francs, but thanks to new industrial partners they were kept at the level before the price increases. Based on customer feedback from 31 breeding boxes, optimizations were implemented, particularly in the removal of insects and the robustness of the boxes. Using image recognition and integration into the IT infrastructure, images in the breeding box can be evaluated automatically and in real The size of the insects, the amount of food, and the movement of the conveyor belt are checked and transmitted. The system is running stably and the failure rate has been reduced from 13% to less than 5%. Customers can see all essential information at a glance on a new display.

2019-2020: In the Erleb-AR project financed by the Swiss Federal Office of Culture an outdoor augmented reality application is developed to visualize 4 cultural heritage sites in Switzerland. Our project is called Erleb-AR, because it is supposed to make the cultural heritage practically tangible with augmented reality. The title Erleb-AR is wordplay with the German adjective “erlebbar” which means in German experienceable and AR.In the project, an app for mobile devices is developed that allows the outdoor AR visualization of historical buildings that don’t exist anymore. To illustrate the project idea, a concept video was created: https://youtu.be/zdezcX9PfRs. As an example of the concept video, we chose the 1834 demolished Christoffelturm on the Berne main station. All those who have installed our app on their mobile device can experience this imposing tower again.

2018-2019: In the ID-Verification (CTI-Nr. 28304.1PFES-ES) 2017-19: Smartphone based software solution for the user verification and the detection of security elements of ID documents.

2017-18: In the ADPS - Alzheimer Disease Prediction Service (KTI 18996.2 PFLS-LS) project, a mobile application is being developed in which information for a non-invasive diagnosis of Alzheimer's is collected using an augmented reality game and the movement sensor data. The game will be played by patients with a risk of Alzheimer Disease. Current state-of-the-art diagnostic measures of Alzheimer's disease (AD) are invasive (cerebrospinal fluid analysis), expensive (neuroimaging) and time-consuming (neuropsychological assessment) and thus have limited accessibility as frontline screening and diagnostic tools for AD. This project proposes a novel computational biomaker for mobile phones (ADPS) which may facilitate early and potentially more effective therapeutic and preventative strategies for AD.

2016-2017: In the Bodygee3D project (CTI 19154.2 PFES-ES), a morphable body model is created using 3D body scans to use 3D body scans in the areas of nutritional advice and fitness. The aim of this project is the development of an analysis module that determines the effectiveness of the habit change/training on the specific body parts based on 3D body scans and enables predictions of body changes using a morphable body model. Obesity is becoming a widespread disease. Only those who permanently change their habits have good chances of reaching and maintaining lower body weight. This project enables not only statements that body fat has been reduced to be made but with much greater accuracy where fat has been lost. In this way, the weight-loss behavior and the effectiveness of the training/behavior change can be documented. In the medium term, the goal is to be able to make additional statements about the relationship between muscle, fat and water content in relation to the individual body parts and by using local impedance measurements. However, this is not part of this project. With the Body Morphable Model, not only can states be recorded, but thanks to the parameterization and the statistically determined data, the course (first legs, then buttocks ...) of the change in body shape can be simulated in the medium term.

2015-2016: In the CleverScan project (KTI 15620-1 PFES-ES), a camera system was developed with which manholes can be scanned and measured in 2D and 3D. The system will go into production in 2016. The new camera produces a detailed 3D-reconstruction of the manhole with detailed images mapped on the 3D surface. The camera uses five sensors: Fore synchronized sensors to the side are used for the scanning and one sensor downwards captures a full HD video stream. The 3D-reconstruction is done using 4 plane lasers.

2014-2015: In the i-Lumica Expert Gate project (KTI 14281-1 PFES-ES), a system consisting of 40 cameras and 20 beamers was developed with which Ha-gel cells can be detected on car bodies. The application realizes several modules to accomplish these tasks. A data acquisition module retrieves 2D and 3D data from the surface scanning hardware provided by i-Lumica AG. Depending on the current analysis run, the 3D data part is cleaned, repaired and optimized to produce mesh data of cars which act as virtual master models. A segmentation module is used to identify, segment and store car body parts based on these master models. Finally, a surface analysis module uses the acquired 2D hail damage data of an ad-hoc car body to simulate hail and assess damaged parts based on the master model of the same car type.
The information gained from such a damage analysis is stored in enhanced databases for further analysis and online access as well as online visualization for 3rd party service providers like insurance companies, logistics experts and engineers.

2012-2014: In the Multimodal Passenger Flow Control System (KTI 13316.1 PFFLE-IW) project, intelligent stereo cameras were developed with which the flow of people in large buildings can be monitored. The camera system was successfully launched on the market.

2015-2017: Virtual Room VR: The Virtual Room project is about allowing several people to meet in a virtual room via the network. Each person wears newer generation VR glasses (e.g. HTC Vive or Oculus Rift) with two hand controllers each, the position and orientation of which are precisely recorded in the respective real room and synchronized via the network. It is possible for a person in Switzerland and a person in Australia to meet in a virtual space and not only communicate verbally and visually, but also interact manually with their hands. Together you can e.g. look at a model of a machine and agree on a possible repair. The respective representation of one's own body in virtual space can be chosen freely. The video below shows a demonstration of the project at the BFH stand at the 2016 professional fair in Bern.

2012: US-Patent: CAMERA-BASED PIPELINE INSPECTION SYSTEM

Janick Stucki Paragliding AR 2023

Bryan Tabinas & Raatheban Rajakumar Product Placement in 3D Environments 2023

David Käser Utility Designer for Unity 2023

Fredy Maeder Visual Inertial Odometry 2023

Marc Affolter Particle System in OpenGL 2022

Damain Schüpbach Sailing VR 2022

Luca Ritz & Lukas Seglias Billiard-AI 2021

Sylvain Barthe & Sandro Scherrrer Goalie VR 2021

Daniel Cameron Automatic Video Object Removal 2020

Olivier Gafner, Frederik Heck, Yannik Stuker Stratoon (Tracked Stratospere Balloon) 2020

David Märki Semantic Segmentation for LOCSIM 2020

Michael Schertenleib Shadow Mapping with OpenGL in SLProject 2020

Steven Henz UI Framework for Oculus VR 2020

Nic Dorner Ray Tracing with Optix 2019

Joel Frutiger HoloLens for Machine Maintenance 2019

Roland Bruggmann Volume Rendering in Unity 2016

Nic Dorner Eularian Video Magnification 2019

Michael Utz Player Tracking 2018

Stefan Töni Voxel Cone Tracing 2018

Jan Dellsperger & David Sheppard Low Cost Motion Capturing System 2018

Matthias Spring VR Kitchen Configurator 2018

Sebastian Häni & Raphael Laubscher Thermal Infrared Scanner for SBB Trains 2017

Vincent Genecand & Lukas Knöpfel Hololens for BIM 2017

Patrik Marti & Mathias Winkler Tetris GameBot 2016

Tina Gerber Face Morphing in PCA Space 2016

Daniel Probst & Kaspar Schmid Billard Tracking at 120 FPS 2013