Klasifikasi tingkat kematangan pisang Cavendish menggunakan convolutional neural network berbasis MobileNetV2

Oktavian, Cindy (2025) Klasifikasi tingkat kematangan pisang Cavendish menggunakan convolutional neural network berbasis MobileNetV2. Sarjana thesis, UIN Sunan Gunung Djati Bandung.

Preview	Text (COVER) 1_cover.pdf Download (166kB) | Preview
Preview	Text (ABSTRAK) 2_abstrak.pdf Download (202kB) | Preview
Preview	Text (PERNYATAAN KARYA SENDIRI) 3_skbebasplagiarism.pdf Download (6MB) | Preview
Preview	Text (DAFTAR ISI) 4_daftarisi.pdf Download (255kB) | Preview
Preview	Text (BAB I) 5_bab1.pdf Download (301kB) | Preview
	Text (BAB II) 6_bab2.pdf Restricted to Registered users only Download (559kB) | Request a copy
	Text (BAB III) 7_bab3.pdf Restricted to Registered users only Download (350kB) | Request a copy
	Text (BAB IV) 8_bab4.pdf Restricted to Registered users only Download (847kB) | Request a copy
	Text (BAB V) 9_bab5.pdf Restricted to Registered users only Download (225kB) | Request a copy
	Text (DAFTAR PUSTAKA) 10_daftarpustaka.pdf Restricted to Registered users only Download (243kB) | Request a copy

Abstract

Cavendish banana is one of the most consumed tropical fruits and is widely sold in markets with varying ripeness levels. The determination of banana ripeness is an important factor that affects market value, taste quality, and consumer appeal. Therefore, an automated system capable of accurately classifying ripeness levels is needed. This study aims to develop a classification system for Cavendish banana ripeness using deep learning methods based on the Convolutional Neural Network (CNN) architecture, specifically MobileNetV2. The banana image dataset was labeled based on the interval of collection days, ranging from Day 0 to Day 9, representing the gradual ripening stages of Cavendish bananas. Image augmentation was applied to balance class distribution and improve model generalization. The model was trained in two epoch scenarios: 10 and 20 epochs. Evaluation results show that training with 20 epochs provided the best performance with an accuracy of 98.7%, along with high precision, recall, and F1-score across all classes. The final model was deployed into a web-based application using the Streamlit framework. The application provides an interactive interface that allows users to upload banana images or use the camera directly to obtain ripeness classification results quickly and easily. With these results, the developed system is expected to assist in the automatic, efficient, and accurate classification of Cavendish banana ripeness.

Actions (login required)

View Item