Development and Testing of Embedded System for Smart Detection and Recognition of Witches’ Broom Disease on Cassava Plants using Enhanced Viola-Jones and Template Matching Algorithm

The Cassava Phytoplasma Disease (CPD) also known as Witches' Broom disease is presently manifesting in some cassava fields. The infection is named after the broom-like leaf spread at the top of cassava plants. This study centered on the development and testing of an embedded system for smart detection and recognition of Witches' Broom disease in natural environment with applied multiple layer validation procedures. The present technology in image processing applying enhanced Viola-Jones and template matching algorithm made the study possible. The embedded system was developed using Python version 3, OpenCV version 3.2.0, and Raspbian Jessie OS. The hardware includes Raspberry Pi 3 Model B and, Raspberry Pi camera module v2 with 8-megapixel camera resolution. A custom cascade classifier model comprising of image acquisition, template selection, train cascade classifier, and real-time object detection and recognition was implemented. With the implementation of two custom trained cascade classifiers, the embedded system was able to lessen the false detection rate compared to the analysis based on general features of a CPD infected cassava leaves alone, thus, increasing its detection accuracy. During the template matching process, 2 voting schemes were used to compare key features from template versus the image of object of interest and used to evaluate the overall results with the 11 templates from the score gauge, thus, these schemes were able to recognize if cassava is CPD infected or not. Currently, the downside of the study is that with the implementation of multiple layers of validation in an attempt to minimize false detection puts too much toll in the micro processing unit which is critical in attempting the detection and recognition in real-time (or close to real-time). Although detection is real time, recognition has a delay in processing the feature of the cassava whether infected or not