Efficient Grading of Tomato Maturity Using a Majority-Vote Committee of Computer-Based Color Classifiers
|
 |
|
Post a Comment
|
|
|
|
|
CONTRIBUTORS:
|
|
|
PROCEEDINGS TITLE:
|
|
|
YEAR:
|
2007
|
|
PUB TYPE:
|
Conference Paper in Proceedings
|
|
PAGES:
|
62 -
n/a
|
|
SUBJECT(S):
|
tomato grading; neural network; computer vision; image processing
|
|
DISCIPLINE:
|
Engineering and Applied Sciences
|
|
HTTP:
|
http://www.ics.uplb.edu.ph/node/255
|
|
LANGUAGE:
|
English
|
|
PUB ID:
|
103-444-130
(Last edited on
2008/07/19 02:27:38 GMT-6)
|
|
SPONSOR(S):
|
|
|
ABSTRACT:
A consumer’s purchase decision of an agricultural produce is usually affected by the product’s color and other external features. The color of tomato is indicative of its ripeness and shelf life. In commercial farms, human graders use the USDA color chart for visual estimation of the tomato’s degree of ripening. The chart helps graders classify the tomato into six ripening stages: Green, Breakers, Turning, Pink, Light Red, and Red. Human graders are naturally subjective and their efficiency in grading diminishes over time due to visual stress and tiredness. Automating the grading process will remove human subjectivity and introduce constant efficiency in grading. Thus, we developed a computer-based vision system using an artificial neural network (ANN) as a maturity classifier. Using a digital camera, we used thousands of color images of human-graded tomatoes to train, test, and validate several ANNs. We picked the ANN with the highest validation rate (97%), comparing its classification with that of the expert graders’. We observed, however, that the 3% wrongly classified tomatoes were correctly classified by most of the ANNs that we did not pick. We further observed that those tomatoes that were wrongly classified by some ANNs were correctly classified by other ANNs. Thus, we selected the ANNs with accuracy rate of at least 90% and combined them into a committee of classifiers, where each member casts a vote as its classification. We take the majority vote as the committee's classification of the tomato maturity. We validated the committee's output over the validation dataset. Results show that the committee agrees 98% of the time with the expert graders, improving the classification efficiency by 1%. The remaining 2% were classified wrongly but within one maturity stage difference.
|
|
|
|
STATISTICS
|
|
Click on # to view
|
|
Citations
|
|
0
|
|
References
|
|
0
|
|
Comments
|
|
0
|
|
Quality
|
|
0/0.00
|
|
Interest
|
|
0/0.00
|
|
View(er)s
|
|
1/108
|
|
|
|
|
|
|
| Prev |
Next |
|