![[Paper] Computer and Software Technology and Robotics (CSR)](../../../../assets/md5images/871014b9248cd5bfeb0a4c8d1f4f51f3.png)
SMARTFruIT: A Deep Learning-Based Convolutional Neural Network for Non-Destructive Watermelon Maturity and Quality Assessment
Document Types
Paper Presentation
Research Theme (for Paper Presentation and Poster Presentation submissions only)
Computer and Software Technology, and Robotics (CSR)
School Name
De La Salle University - Laguna Campus
Track or Strand
Science, Technology, Engineering, and Mathematics (STEM)
Research Advisor (Last Name, First Name, Middle Initial)
Ramos, Christian Daniel, E.
Start Date
25-6-2026 10:30 AM
End Date
25-6-2026 12:00 PM
Zoom Link/ Room Assignment
DLSU Laguna Campus (In-person) - Enrique K. Razon Jr. Hall - EKR 404
Abstract/Executive Summary
Agriculture remains a critical sector in the Philippines, yet postharvest losses, particularly during grading and handling, continue to reduce product quality and farmer income. Traditional fruit grading methods rely heavily on manual inspection, which is often subjective and inconsistent. This study aims to address this issue by developing SMARTFruIT, a low-cost, artificial intelligence (AI)-based system designed to classify watermelon maturity using image-based analysis. Specifically, the study investigates whether a Convolutional Neural Network (CNN) can accurately distinguish between ripe and unripe watermelons using smartphone-captured images.
A quantitative research design was employed, utilizing a pre-labeled dataset of watermelon images. The images were processed using the HSV color space to enhance color feature detection, and a CNN model was developed using the PyTorch framework. The dataset was divided into training, validation, and testing sets, and the model was trained over multiple epochs to learn distinguishing visual patterns such as color, texture, and surface characteristics.
Results show that the SMARTFruIT system achieved a high classification accuracy of 97.83%, demonstrating its effectiveness in identifying watermelon maturity. While some limitations were observed, particularly in cases with subtle visual differences or varying lighting conditions, the model maintained overall stable performance.
The findings support the claim that low-cost AI-based systems can provide a reliable and accessible alternative to traditional fruit grading methods. SMARTFruIT has the potential to improve grading consistency, reduce human error, and contribute to more efficient and sustainable agricultural practices.
Keywords
artificial intelligence; fruit maturity classification; convolutional neural networks; image-based analysis; smart agriculture
Initial Consent for Publication
yes
Statement of Originality
yes
SMARTFruIT: A Deep Learning-Based Convolutional Neural Network for Non-Destructive Watermelon Maturity and Quality Assessment
Agriculture remains a critical sector in the Philippines, yet postharvest losses, particularly during grading and handling, continue to reduce product quality and farmer income. Traditional fruit grading methods rely heavily on manual inspection, which is often subjective and inconsistent. This study aims to address this issue by developing SMARTFruIT, a low-cost, artificial intelligence (AI)-based system designed to classify watermelon maturity using image-based analysis. Specifically, the study investigates whether a Convolutional Neural Network (CNN) can accurately distinguish between ripe and unripe watermelons using smartphone-captured images.
A quantitative research design was employed, utilizing a pre-labeled dataset of watermelon images. The images were processed using the HSV color space to enhance color feature detection, and a CNN model was developed using the PyTorch framework. The dataset was divided into training, validation, and testing sets, and the model was trained over multiple epochs to learn distinguishing visual patterns such as color, texture, and surface characteristics.
Results show that the SMARTFruIT system achieved a high classification accuracy of 97.83%, demonstrating its effectiveness in identifying watermelon maturity. While some limitations were observed, particularly in cases with subtle visual differences or varying lighting conditions, the model maintained overall stable performance.
The findings support the claim that low-cost AI-based systems can provide a reliable and accessible alternative to traditional fruit grading methods. SMARTFruIT has the potential to improve grading consistency, reduce human error, and contribute to more efficient and sustainable agricultural practices.
https://animorepository.dlsu.edu.ph/conf_shsrescon/2026/BoA_CSR/16
