DETEKSI KEMATANGAN BUAH MANGGA AMPLEM SARI MENGGUNAKAN ALGORITMA CONVOLUTIONAL NEURAL NETWORK

Handayani, Ni Made Dewi and Karmila, Sely and Rusjdi, Darma (2024) DETEKSI KEMATANGAN BUAH MANGGA AMPLEM SARI MENGGUNAKAN ALGORITMA CONVOLUTIONAL NEURAL NETWORK. Diploma thesis, ITPLN.

[thumbnail of 202031134_NI MADE DEWI HANDAYANI_REVISI_SKRIP_NI MADE DEWI Handaya.pdf] Text
202031134_NI MADE DEWI HANDAYANI_REVISI_SKRIP_NI MADE DEWI Handaya.pdf
Restricted to Repository staff only

Download (4MB)

Abstract

Penentuan kematangan buah secara akurat sangat krusial dalam menjaga kualitas produk pertanian. Metode konvensional yang mengandalkan penilaian manual seringkali tidak efisien dan rentan kesalahan. Hal ini menjadi kendala dalam menjaga kualitas produk pertanian. Oleh karena itu, penggunaan metode otomatis untuk mendeteksi kematangan buah menjadi semakin penting. Memanfaatkan keberhasilan Convolutional Neural Network (CNN) dalam klasifikasi citra, pendekatan pembelajaran mendalam diterapkan untuk mendeteksi kematangan buah mangga Amplem Sari berdasarkan warna kulitnya. Dataset terdiri dari 951 citra buah mangga yang dikategorikan menjadi empat kelas yaitu belum matang, mulai matang, setengah matang, dan matang sempurna. Berbagai arsitektur model CNN digunakan, termasuk Simple CNN, VGG16, Xception, dan DenseNet121. Kinerja setiap model diuji menggunakan data uji yang belum pernah dilihat sebelumnya dan diukur melalui metrik akurasi, presisi, recall, dan F1-score. Hasil penelitian menunjukkan bahwa model DenseNet121 memiliki kinerja terbaik dengan akurasi 92.63%, diikuti oleh VGG16 sebesar 90.53%, Simple CNN sebesar 87.37%, dan Xception sebesar 82.11%.

Accurate determination of fruit ripeness is crucial in maintaining the quality of agricultural products. Conventional methods that rely on manual assessment are often inefficient and error prone. This is an obstacle to maintaining the quality of agricultural products. Therefore, the use of automated methods to detect fruit ripeness is becoming increasingly important. Leveraging the success of Convolutional Neural Network (CNN) in image classification, a deep learning approach is applied to detect the ripeness of Amplem Sari mango fruit based on its skin color. The dataset consists of 951 mango fruit images categorized into four classes: not yet ripe, starting to ripen, half ripe, and fully ripe. Various CNN model architectures were used, including Simple CNN, VGG16, Xception, and DenseNet121. The performance of each model was tested using never before-seen test data and measured through accuracy, precision, recall, and F1-score metrics. The results show that the DenseNet121 model has the best performance with 92.63% accuracy, followed by VGG16 at 90.53%, Simple CNN at 87.37%, and Xception at 82.11%.

Item Type: Thesis (Diploma)
Uncontrolled Keywords: Convolutional Neural Network, Kematangan Buah, Klasifikasi, Mangga Amplem Sari, Pembelajaran Mendalam Convolutional Neural Network, Fruit Ripeness, Classification, Mango Ample Sari, Deep Learning
Subjects: Skripsi
Bidang Keilmuan > Teknik Informatika
Divisions: Fakultas Telematika Energi > S1 Teknik Informatika
Depositing User: Sudarman
Date Deposited: 15 Sep 2025 03:55
Last Modified: 15 Sep 2025 03:55
URI: https://repository.itpln.ac.id/id/eprint/1057

Actions (login required)

View Item
View Item