Show all
Show all
Title Description Supervisors Status
Design and optimization of an electroacoustic absorber using multiple resonators The electroacoustic absorber is an electroacoustic device for controlling the acoustic impedance of a closed-box electrodynamic loudspeaker diaphragm, in order … Etienne Rivet Finished
Electroacoustic absorption and Energy harvesting The electroacoustic absorber is an electroacoustic device allowing the control of the acoustic impedance of a loudspeaker diaphragm, in order … Etienne Rivet Finished
Binaural rendering of 3D audio environments on smartphones The rendering of 3D audio environments can be achieved by many means, such as 2D or 3D loudspeaker arrays, or … Sami Karkar, Gilles André Courtois Finished
Psychoacoustic comparison of 3D audio rendering In fall 2016, the new building « Under One Roof » will open on the EPFL campus. In this building, … Finished
Developpement of an active noise control umbrella SounDDot is a technology recently developed in the laboratory, providing highly directive sound-diffusion. It is based on an array of … Romain Boulandet, Thomas Laurence Finished
MNIST classification using deep learning This project consists in a smooth introduction to machine learning and particularly deep learning techniques. The student will first be … Nauman Shahid, Nathanaël Perraudin Finished
Creation and analysis of large social networks, an experiment on the Star Wars universe The goal of this project is to build a social network of the characters appearing in a finite corpus of … Kirell Benzi, Benjamin Ricaud Finished
Manifold frame bounds for graph frames The goal of this project is to examinate the properties of a new objet that extends a graph kernel (or … Johann Paratte, Nathanaël Perraudin, Daniel Kressner Finished
Audio Steganography using Convex Demixing ** Idea :** design and implement a method to convey information on an audio stream using steganography with the use … Johann Paratte Finished
Robust Visual Tracking Using Feature Selection Idea : provide a complete survey of the state-of-the-art algorithms for visual tracking and improve the best-performing algorithm using MLBP … Johann Paratte Finished
Towards Sparse Scene Geometry Recovery Only Using Mobile Device Idea : recover the geometry of a scene using a mobile device camera with online on-device computation only. Objectives : … Johann Paratte Finished
Multi-view Stereo for 3D reconstruction ** Idea :** use state-of-the-art projective geometry techniques to resolve the structure from motion problem and spatial calibration of cameras. … Johann Paratte Finished
3D Scene Reconstruction from a Single Movie Idea : provide a scheme to automatically infer the 3D model of the static background in a scene from a … Johann Paratte Finished
Graph-based Image Inpainting ** Idea :** use a patch-based graph of the image to address the inpainting problem for large missing chunks. Objectives … Johann Paratte, Nathanaël Perraudin Finished
Realtime Face Detection and Tracking on Mobile for Anonymization and Privacy Idea : provide selective anonymization of people on video and photographs using face detection and tracking on mobile devices. Objectives … Johann Paratte Finished
Stochastic processes on graphs In this project, we study how stochastic processes behave on graph. The main idea is to generate many random signals … Nathanaël Perraudin, Olivier Lévêque Finished
Movie Summary Using Dimensionality Reduction Idea : provide a meaningful visual summary of a video sequence in an entirely automatic way. ** Objectives : ** … Johann Paratte Finished
Distance learning for smooth graph signals In this project, we study a new technique for distance learning on a graph. Traditional distance learning minimizes the distance … Vassilis Kalofolias, Nathanaël Perraudin Finished
Learning with Hypergraphs In this project the student will inquiries hypergraphs and generalizes the spectral graph theory to this setting. This project is … Nathanaël Perraudin Finished
Robust Face Detection and Recognition using Depth Cameras Idea : achieve face recognition based on RGB-D cameras. ** Objectives :** creation of a 3D dataset for the learning … Johann Paratte Finished
Object Classification using Depth Cameras ** Idea : ** tackle object classification tasks based on depth sensor input. ** Objectives : ** finding a way … Johann Paratte Finished
People Recognition in the Montreux Jazz Festival Archive ** Idea :** Provide an automatic labelling of people in videos using face detection and visual tracking, with application to … Johann Paratte Finished