Classification of thrips-infected chili plant using image processing on embedded system

Al-Maswari, Majdadeen Hazaa Saad (2021) Classification of thrips-infected chili plant using image processing on embedded system. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

	Text (24 Pages) Classification of thrips-infected chili plant using image processing on embedded system.pdf - Submitted Version Download (719kB)
	Text (Full text) Classification of thrips-infected chili plant using image processing on embedded system.pdf - Submitted Version Restricted to Repository staff only Download (2MB)

Abstract

In chili plantations, infestation by insects bacterial, and fungal diseases are the major constraints in chili production. The infestation by thrips for example can cause curling on the leaves which can be monitored remotely using cameras. Thus, this project will embark on classifying the leaves infested by thrips from healthy leaves. Previous works have shown the possibility of using texture features, shape features and various combinations of K-means clustering algorithms to highlight the pathological differences. Images were captured using the camera and processed onboard the Raspberry Pi. The images were processed with the help of Open-CV libraries and in the Python environment. Texture features and shape features were computed from the processed images before it is fed to Linear Support Vector Machine algorithm (SVM) for supervised learning. The output model was used to test images for accurate classification. The results were analyzed in the form of a confusion matrix. So, the Linear SVM classifier shows an accuracy of 80%, specificity of 75%, precision of 78% and Sensitivity of 84%.

Actions (login required)

View Item