Design and development of eggplant fruit and shoot borer (leucinodes orbonalis guenee) detector using near-infrared spectroscopy / by Maria Patrice L. Lajom and Joseph Paul R. Remigio.

Design Project (Bachelor of Science in Electronics and Communications Engineering) Cavite State University.

Includes bibliographical references.

College of Engineering and Information Technology (CEIT)

LAJOM, MARIA PATRICE L., REMIGIO, JOSEPH PAUL R. Design and Development
of Eggplant Fruit and Shoot Borer (Leucinodes orbonalis Guenée) Detector Using Near-
Infrared Spectroscopy. Undergraduate Project Design. Bachelor of Science in Electronics
Engineering. Cavite State University, Indang Cavite. December 2022. Adviser: Edwin R.
Arboleda, DEng.
The Eggplant Fruit and Shoot Borer (EFSB) detector was designed and developed to
non-invasively classify eggplant fruits that are non-infested and infested with EFSB. The study
aims to improve the traditional process of EFSB classification within an eggplant fruit through
the use of the NIRS technique. NIRS allows on-line and non-invasive monitoring of a fruit's
internal quality by detecting changes in its chemical and optical properties due to insect
infestation. Relevant information is therefore gathered through quantifying the interaction
(energy attenuation) between the IR energy and the food samples. The study used Support
Vector Machine as its machine learning classifier algorithm.
Using kernel methods, SVMs effectively carries out non- linear classification by implicitly
projecting their inputs into high-dimensional feature spaces. Since the prototype was developed
for on-line monitoring, portability was given of utmost importance, patterning the design in the
form of a handheld gun. The 3D-printed PLA chassis houses its components: Arduino Nano,
AS7263 Spectral Sensor, OLED Display, MicroSD Card module, power supply, and switches.
The classifier model was trained and developed using Jupiter and was extracted as a C++ code
for Arduino. Through this, the model was successfully deployed to the Arduino microcontroller.
The prototype's performance was compared to the farmer's traditional performance of
classification in terms of accuracy, precision, and speed. The findings indicated that the
prototype performs better in all three aspects, with an accuracy of 84%, precision of 72.83%,
and an average speed of 9.736 seconds.

Submitted to the University Library 08/31/2022 DP-752