Labworks in Computer Vision
TP1: Panorama Stitching
- Subject page
- My sources along with some pictures you can use
TP2: Camera Calibration
TP3: Structured Light
- Subject page
- My sources along with a report in .pdf and some pictures to play with
TP4: Stereo
- Subject page
- My sources along with a report in .pdf
Unsupervised Learning with Theano
During my winter spent in the Université de Montréal in Québec, I worked on a collective project with my classmates from the course algorithmes d’apprentissage. The long-term objective of the project was to develop new facilities to ease unsupervised learning tasks in python with the help of the Theano library, while our short-term motivation was to take part of the Unsupervised and Transfert Learning Challenge. Our team (LISA) did a honorable score in the contest, with a rather fruitful tool development. The online repository of our toolbox can be found here on Assembla.
On the Road Coloring Problem
For our evaluation during the the M2 course Recent Topics in Graph Theory, we had to study a specific issue and realize a small survey on the matter, by summing up several research papers and detailling on or two technical points along the way. I devoted myself to the road coloring problem, which deals with edge coloring of a graph, the existence of a synchronizing word in a finite automaton, and cycles in a graph of relativeley prime lengths. You can find my report here, as well as the slides used for my oral presentation.
Parallel Oceanic Simulation with MPI
Part of the evaluation of our M1 course in Parallel Algorithm consisted in a small oceanic simulation project using MPI. The problem consists in a 2d grid populated by sharks and sardines, the evolution of which is regulated by local rules. The grid is then splitted into horizontal bands, each of them being treated in parallel in order to gain in efficiency compared to a sequential algorithm. The detailed subject can be found here (page 1 and page 2), along with my report and my source code.