Automated Rose Maturity Stage Detection

From Master Projects
Jump to: navigation, search


has title::Automatic Rose Grading
status: ongoing
Master: project within::Computational Intelligence and Selforganisation
Student name: student name::Chris van den Berg
Dates
Start start date:=2011/02/01
End end date:=2011/07/01
Supervision
Supervisor: Wojtek Kowalczyk
Company: has company::Logica
Poster: has poster::Media:Media:Posternaam.pdf

Signature supervisor



..................................

Abstract

In the flower industry flowers sold for esthetical purposes are graded on a variety of features such as length or defects. One of these features is the maturity stage, which depicts the extent to which the bud develops during the flowers vase life. The Association of Dutch Flower Auctions (VBN) makes a distinction between seven different stages of maturity, from which five are qualified for trading. A maturity of stage one means the flower is still very coarse (it is even possible the flower will experience great difficulty opening up) and a flower at maturity stage five is opened and the colours have fully developed. After sorting, flowers are grouped by their maturity stage, auctioned and distributed. Currently the classification of the maturity stage for most flowers is done manually by visual inspection. This can be tedious and error prone task, because the inter class differences are small and different persons might have a different method of classification. A lot of research has been done in the field of face recognition, as well as image classification and image based food qualification. From an image analysis point of view, the problem can be seen analogous to automated age detection in humans, which is similar in the sense that the set of class labels is a totally ordered set, each ripeness has a unique rank in the time sequence. The goal of this project is to investigate if the techniques from those fields are applicable to the task of assigning the correct maturity stage to a flower.