A Framework for Grading of White Chali Type Arecanuts with Machine Learning Algorithms
Kusumadhara S1, Ravikumar M S2, Raghavendra P3

1Kusumadhara S, Department of Electronics and Communication Engineering, K V G College of Engineering, Sullia and affiliated to Visvesvaraya Technological University, Belagavi, Karnataka, India.
2Ravikumar M S, Department of Electronics and Communication Engineering, K V G College of Engineering, Sullia and affiliated to Visvesvaraya Technological University, Belagavi, Karnataka, India.
3Raghavendra P, Senior Manager, CAMPCO Ltd., Sullia, South Canara District, Karnataka, India.
Manuscript received on March 12, 2020. | Revised Manuscript received on March 25, 2020. | Manuscript published on March 30, 2020. | PP: 2782-2788 | Volume-8 Issue-6, March 2020. | Retrieval Number: F8389038620/2020©BEIESP | DOI: 10.35940/ijrte.F8389.038620

Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Grading of arecanuts before marketing fetches better prize. There are two universally accepted grading systems for arecanuts namely grading at the producers’ level and grading at the wholesale traders’ level. The major work in this paper is devoted to generation of a standard image database for the White Chali Type arecanuts and constructing a frame work for their grading by exploring the features of White Chali Type arecanut images for the first time. Further, two separate datasets are developed for the above grading systems by employing image feature extraction methods with 3500 and 4132 instances respectively. The arecanuts are graded using popular machine learning algorithms and the results are validated using ten fold cross validation. Multinomial logistic regression as classifier outperformed with classification accuracies of 98.8% and 92.69% for the producers’ level and the whole sale traders’ level grading systems respectively. Also the performances of various machine learning algorithms for the above two datasets are evaluated using weighted average values of True Positive rate, False Positive rate, precision, recall, F-measure and Cohen’s Kappa coefficient.
Keywords: Arecanut Grading, Cohen’s Kappa Coefficient, Cost Sensitive Classification, Machine Learning.
Scope of the Article: Machine Learning.