Vincent Vuarnoz

MSc Semester Project

Title

Flower recognition

Candidate

Vincent Vuarnoz

Supervisor

Prof. Dr. Touradj Ebrahimi

Assistant: Peter Vajda

Expected

June 11, 2010

Place

EPFL

Description

Imagine you are hiking in the Swiss mountains with your friends. You take a picture of a nice flower with your mobile phone, but you do not know what is the name of this flower and what are its main features. Maybe you found a protected, very rare flower! In order to have some information about the flower, you could send its picture to a server, so that it can show on your mobile phone some information about it. The application can be interactive and can use the environment variables such as GPS data or date.
The goal of this project is to develop a flower recognition application based on special visual features of such objects. These features are based on the real features that biologists would use for flower classification. The following tasks should be performed:

  1. Review the state of the art in flower recognition.
  2. Study biological point of view of flower recognition.
    1. Meet and discuss the task with a professional scientist in this field.
  3. Collect or create a small flower database.
  4. Implement and validate a mobile client application on an Android phone, containing the following parts:
    1. GUI for mobile phone, which allows capture and display of a flower.
    2. Communication module through a wireless channel to exchange data/image with a server.
  5. Implement and validate a server side application. This application can use existing methods and do not need to be re-implemented all from scratch. It should contain:
    1. Communication framework with mobile through wireless channel.
    2. Modular and extensible framework for feature extractions.
    3. Feature extractions with distance measurement including the following features categories: Color features, Shape features.
    4. Segmentation.
    5. Database search.
  7. Assessment and comparison of different features and their respective distance metrics.
  8. Demonstration application on mobile phone in such a way that it can be used by native users.