Journal of Multi Disciplinary Engineering Technologies, Vol. 14, Issue 1

Paper Title: Design of an Automatic Image Detection Algorithm for On-tree Green Citrus Fruit

Authors: Jigar A. Soni, Hetal N. Patel

Corresponding Author Email: jasoni280489@gmail.com

Abstract: An automatic on-tree green fruit detection algorithm developed using Machine learning. It is difficult to differentiate green citrus from its background due to the correlation in the color of leaves and fruit. The detection is done using histograms of oriented gradients (HOG), Local binary patterns (LBP), Haar & Support Vector Machine on the training of 2100 on-tree green fruit images. The results show that the proposed approach is capable of automatically detecting the green citrus fruit with a high degree of accuracy. The results can optimize by proper selection of several stages and false-positive rates in the training process. The results are compared with ground truth data. The proposed algorithm is suitable for yield measurement/ monitoring analysis of crops for agriculture applications.

PDF Link