2015 Grooming app: The Philips grooming app can give photo-realistic beard style previews. Task: convert Matlab research code into C++ production code. (freelance project)
2014 ‘Virtual Dressing Room’: A customer can virtual fit clothes standing in front of a ‘mirror’
2013-2014 ‘3D Photo Viewer’ at Start-up Auto3D: I developed an structure from motion algorithm that can obtain a photo-realistic 3D model from a large unstructured set of 2D images. The 3D model can be used in the 3D photo viewer to navigate through a large set of photo’s. I developed the structure from motion in C++ and used a MySQL-database to store the raw data and intermediate results. To visualize the intermediate results, I used OpenGL and interfaces made with Qt. For the 3D photo viewer, I first made a prptotype in JavaScript and later on I used Unity3D and C#. The large scale matching and retrieval of the images was done with machine learning, in Java. An example of an object captured by the tool can be found at www.Auto3D.com.
2013 Analysis of endoscopic images with computer vision techniques at Philips Research. Endoscopic video is captured in the throat of a person that is suffering from sleep apnea. I wrote code to analyze the visual data in Java with the ImageJ API and in C++ with OpenCV. (freelance project)
2011-2013 ‘Vital Sign Camera’: An app was added to fitness equipment that can determine the heart rate of a exercising person by looking at that person with a normal camera. I was involved during the research phase (Matlab) and during industrialization (C++).
2012 Domain name catch app: I build a small Android app to catch expired domains.
2012 ‘Oled Wall’: I added real-time face tracking and visual effects to the interactive Oled Wall (C++)
2009-2011 ‘Autonomous vacuum cleaner’: I developed a visual-SLAM (Simultaneous Localization and Mapping) algorithm for an autonomous household robot in Matlab. The robotics team used Scrum during developed.
2010 Steering an Actual Size Jeep with a Wii Remote: A weekend nerd project, creating a system with which we could remotely drive and steer the TNO RoboJeep
2009 3D object recognition by means of ToF cameras (Time of Flight) and stereo cameras.
2008-2009 Self-calibration of a stereo rig. A stereo camera of a robotic vacuum cleaner is calibrated automatically when it is moving through a partially static environment. By making automatic re-calibration possible, the cost of the stereo camera setup could be reduced. I designed the calibration algorithm in Matlab.
2007-2008 MSc-thesis at TNO; SLAM (Simultaneous Localization and Mapping) based on SIFT (Scale Invariant Feature Transform). I designed the SLAM algorithm in Matlab and I made a 3D viewer in Java and OpenGL to visualize the 3D point cloud.
2005 2nd place at the Robochallenge. I constructed a robot with omnidirectional steering and wrote the software for perceiving the environment and steering the robot in Java and C++. The goal of the Robochallenge contest is to capture colored balls that are hanging in a delimited game area. Part of the calculations are taking place on the robot itself but the time consuming computer vision calculations and the calculations for the Kalman filter that is used for localization, are handled by 3 computers near the sideline of the playing field by using remote method invocation over a wireless network.

[site is under construction]