AlexandreAlahi

Contents

Home
Research Projects
Publications
Teaching
Data set
Source Code
CV
Media Coverage


Research fields

Computer Vision
Signal Processing
Machine learning
Robotics
Computational Neuroscience


Research Interests

Sparse approximation
Compressed-sensing
Inverse problems
Real-time vision
Bio-inspired vision
Large-scale vision
Big Visual data

 

Contact me if you are interested to work on the following projects:

  1. Image-based retrieval: search within images
  2. Audio similarity in the compressed domain

Given the new microsoft Kinect camera, we also propose the following projects:

  1. Self-calibration of several depth cameras
  2. Background modeling in the presence of people
  3. Gesture recognition
  4. A virtual golf teacher
  5. Face recognition in dark

Image-based Retrieval: Search Within Images

Type: Semester or Diploma project

Image retrieval Google offers a text search engine. The goal of this project is to offer an image search engine. Given a single image of an object, a search engine is developed to find similar objects wihin a database of images.

Keywords: image-based retrieval, interest point, historgram of gradients
Assistant: Alexandre Alahi
Requirements: C/C++
Contact: alexandre.alahi@epfl.ch, phone 35652, ELE 238
Group: Prof. Pierre Vandergheynst


Audio Similarity in the Compressed Domain

Type: Semester or Diploma project

GeneZik Computing the audio similarity between two songs can be useful to either recommend musics or to automatically create playlists without the use of any tags. The goal of this project is to evaluate a set of algorithms that best estimate the audio similarity among a collection of songs in an efficient way, i.e. without un-compressing the mp3 format.

Keywords: Audio, signal processing
Assistant: Alexandre Alahi
Requirements: C/C++
Contact: alexandre.alahi@epfl.ch, phone 35652, ELE 238


Self-Calibration of Several Depth Cameras

Type: Semester or Diploma project

kinect detecting people We developped a system to detect and track people with several Kinect cameras. In order to enable cameras to collaborate, they need to be calibrated. Current calibration step requieres to put some markers in the scene. The goal of this project is to calibrate cameras without any markers. Given the depth captured by the cameras, the cameras should detect the ground plane and similar object to auto calibrate themselves.

Keywords: Multi-view, calibration, depth camera, optimization
Assistant: Alexandre Alahi
Requirements: C/C++
Contact: alexandre.alahi@epfl.ch, phone 35652, ELE 238


Background Modeling in the Presence of People

Type: Semester or Diploma project

Background modeling To detect moving people in a single fixed camera, one crucial step is to extract foreground silhoeuttes with a background subtraction algorithm. In order to model the background, statistics are computed when people are not present in the scene. However, it is not feasable to assume that we can always require to monitor an empty scene specially in urban area. The goal of this project is to model the background even if moving people are present in the scene.

Keywords: Background subtraction, foreground extraction, statistic
Assistant: Alexandre Alahi
Requirements: C/C++
Contact: alexandre.alahi@epfl.ch, phone 35652, ELE 238
Group: Prof. Pierre Vandergheynst


Gesture Recognition

Type: Semester or Diploma project

Gesture Recognition A natural way of human-computer interaction is through gestures. Microsoft has recently offered to naviguate within the Xbox interface with natural gestures given the depth captured by their Kinect camera. The goal of this project is to develop similar algorithm to recognize natural gesture to naviguate within a database of media content.

Keywords: Gesture recognition, depth camera, optical flow
Assistant: Alexandre Alahi
Requirements: C/C++
Contact: alexandre.alahi@epfl.ch, phone 35652, ELE 238
Group: Prof. Pierre Vandergheynst


A Virtual Golf Teacher

Type: Semester or Diploma project

Golf The aim of this project is to bring a personal instructor in every home.  It is proposed to develop a virtual instructor to teach and enhance human gestures. The potential applications can be in arts as well as in sports.
The trainee will perform the gestures of interest in front of a Kinect Camera and the Virtual Instructor (Artificial Intelligence) will give feedbacks. The first application will be to teach golf gestures.

Keywords: Virtual instructor , gesture training system.
Requirements: C/C++
Contact: alexandre.alahi@epfl.ch, phone 35652, ELE 238
Group: Prof. Pierre Vandergheynst


Face Recogntion in Dark

Type: Semester or Diploma project

Face Recognition Face recognition is widely used to enter secured sites. However, most of the algorithms are sensitive to the scene illumination and lighting conditions. The goal of this project is to recognize face regardless the lighting conditions. The noisy depth information captured by the Kinect will be used. The student will evaluate the performance of noisy 3D models to extract features and classify them.

Keywords: Face Recognition, 3D model, classification, feature extraction
Requirements: C/C++
Contact: alexandre.alahi@epfl.ch, phone 35652, ELE 238
Group: Prof. Pascal Frossard

top

© alahi {at} stanford.edu
updated: January 2015

Follow me on

Alahi on ScholarAlahi on LinkedinAlahi on Twitter

 

Schools Studied


Stanford


Ecole Polytechnique Fédérale de Lausanne


Labs Worked

Vision lab at Stanford

LTS2

Transpor


Companies Worked

VisioSafe