Towards Smart Farming Approach for Plants Disease Detection and Classification by Using HoG Feature Extraction and PNN Algorithm

juliet kadum

Computer Science Department ,College of Science/University of Diyala

Jamal Mustafa Al-Tuwaijar

Amaal Kadum

Sadam Hameed Rasheed

DOI: https://doi.org/10.47831/mjpas.v3i1.121

Keywords: 2D Log Chromaticity Image, PNN Machine Learning, HoG Feature Extraction, Evaluation Measures.


Abstract

The advancement of computerized automated diagnostic technologies ensures that health plants are more effective by detecting diseases early. The damage caused by them can be minimized. Recently, many scientists have been trying to use methods and algorithms in image processing and extracting features based on color, texture, and shape and employing them in building smart systems capable of detecting plant diseases in their early stages. In this paper, proposed a smart farming aims to automated remove shadow and segmentation the pepper fruit diseases region, then extracted strong features from diseases region, and finally detected and diagnosis pepper fruit disease using PNN algorithm. In addition, in this work, (Diyala Pepper), which relies on a dataset of pepper plants named Diyala Pepper. The suggested system's implementation was done on a real dataset of pepper fruits. Experimental results of the proposed method showed 100% accuracy in the training phase and 81.82% in the phase of testing.