Minimalist AdaBoost for blemish identification in potatoes

Barnes, Michael and Cielniak, Greg and Duckett, Tom (2010) Minimalist AdaBoost for blemish identification in potatoes. In: International Conference on Computer Vision and Graphics 2010, Sep 20-22 2010, PJIIT - Polish-Japanese Institute of Information Technology.

WarningThere is a more recent version of this item available.

Documents
Minimalist AdaBoost for blemish identication in potatoes
The original document is available via www.springerlink.com
[img]
[Download]
[img]
Preview
PDF
barnes10multiclass.pdf - Whole Document

1MB

Official URL: http://www.springerlink.com/content/1871m28r0362l7...

Abstract

We present a multi-class solution based on minimalist Ad-
aBoost for identifying blemishes present in visual images of potatoes.
Using training examples we use Real AdaBoost to rst reduce the fea-
ture set by selecting ve features for each class, then train binary clas-
siers for each class, classifying each testing example according to the
binary classier with the highest certainty. Against hand-drawn ground
truth data we achieve a pixel match of 83% accuracy in white potatoes
and 82% in red potatoes. For the task of identifying which blemishes
are present in each potato within typical industry dened criteria (10%
coverage) we achieve accuracy rates of 93% and 94%, respectively.

Item Type:Conference or Workshop Item (Paper)
Additional Information:We present a multi-class solution based on minimalist Ad- aBoost for identifying blemishes present in visual images of potatoes. Using training examples we use Real AdaBoost to rst reduce the fea- ture set by selecting ve features for each class, then train binary clas- siers for each class, classifying each testing example according to the binary classier with the highest certainty. Against hand-drawn ground truth data we achieve a pixel match of 83% accuracy in white potatoes and 82% in red potatoes. For the task of identifying which blemishes are present in each potato within typical industry dened criteria (10% coverage) we achieve accuracy rates of 93% and 94%, respectively.
Keywords:Potatoes, Computer Vision, Machine Learning, bmjtype, bmjanomaly
Subjects:G Mathematical and Computer Sciences > G760 Machine Learning
G Mathematical and Computer Sciences > G740 Computer Vision
Divisions:College of Science > School of Computer Science
ID Code:3885
Deposited By: Michael Barnes
Deposited On:19 Jan 2011 14:01
Last Modified:13 Mar 2013 08:53

Available Versions of this Item

Repository Staff Only: item control page