Mohamed Ikbel Boulabiar

Aptum is focusing on automating tasks using new computer vision techniques.

Concepta.me is a web application aiming to change the status quo of creating animated content on the internet.

I worked on : - The development of a facial recognition system. - Augmented Reality application to show information on top of 3D models - Various Web Development - Automatic and wireless recuperation of photos from a DSLR and cloud transfer

I worked on: - Writing internal research documents. - Project proposals, research meetings - Various meetings preparation and summarization with clients - Preparing a scientific article

First program for accelerating startups. The program has marketing, business model and pitch training among other trainings.

Gesture Recognition, - I worked on creating a 3D gesture recognizer in C++ directly using the point-cloud provided by the Kinect (using libfreenect). - I was able to detect pointing in space and on the table along with other hand gestures including pinch. - I created a demonstrator using javascript with the WebGL library Three.js which is connected to the C++ gesture recognizer using a websocket connection. Suspicious ship tracking system (Reconsurve project), - I participated in the specification and development of the project and defining tracking algorithms. - I created a demonstrator using C++ and Qt which included a web based S57 map system. - I worked on connected C++ code with the webview map using javascript injection and input events duplication. - I integrated ship information from a PostgreSQL database delivering XML data. - I worked on animating ship movements from time-series events. Dynamic GUI application Piping - I've used the DBus software bus on Linux to be able to dynamically link GUI application and pipe them in a similar way to what is used for command based applications. - I've used C++ and Qt to create prototypes and demonstrate the concept. 3D pointing and command system (Geste3D project) - In a project funded by Thales, I participated in the development of commanding and manipulating windows in a room. This is based on the gesture recognizer I created in my PhD work. The demonstrator used mainly the pointing by index finger to move windows in the room with a dynamic and intuitive coordinates system.

Multi-touch team. I participated in the development of the uTouch software stack. I participated in the development of the recognition algorithm and the discussion about the architecture. I have released the GINN daemon allowing applications to support multitouch without rewriting. The daemon subscribes to the gestures and according to the active application, it injects keyboard shortcuts allowing the user to feel as they are already supporting multi-touch. (GINN is still installable in ubuntu via apt install ginn)

Part-time job in the Multimedia Lab. Test, debug and parameter a beta "Scalable Video Coding" Codec from Mainconcept Company. - I've proposed and developed a GUI with C++ and Qt to simplify mainconcept's codec parametrization. - I've used DirectShow tool GraphEdit to test and prototype multimedia pipelines with different parameters. - I've made a comparison of available codecs and the different format and containers.

Internship in "Interactive Computing Lab" (laboratoire d'informatique interactive) - I discussed on a daily basis and written a state-of-the-art of input techniques. - I learned and participated in the reverse-engineering of low-level multi-touch events sent through USB using the HID protocol. - I participated in the creation of the first multi-touch protocol in the Linux Kernel (Protocol A) with a team of international collaborators including mainly Stéphane Chatty, Henrik Rydberg and Peter Hutterer. - I created a small gesture recognizer in C99, I also used DBus software bus. - I proposed and created a video demonstration of multi-touch use which gathered more than 145K+ views on Youtube. - I worked on generating input events in a small prototype using the uinput kernel module to allow new forms of input in applications.

I worked on optimizing SystemC microchip simulation. (SystemC is a C++ library) - I used GCC tools on linux (and UNIX) including profilers and optimizers (gperf, gcov, ..) to detect functions taking too much time and unused functions to delete. - I was able to optimize workloads ranging from 2% to 40% (one module).