IMPLEMENTASI TRANSFER LEARNING DENGAN PERBANDINGAN NILAI LEARNING RATE PADA KLASIFIKASI CITRA PENYAKIT DAUN TEBU BERBASIS WEB

Firmansyah, Syafri (2025) IMPLEMENTASI TRANSFER LEARNING DENGAN PERBANDINGAN NILAI LEARNING RATE PADA KLASIFIKASI CITRA PENYAKIT DAUN TEBU BERBASIS WEB. Undergraduate thesis, UPN "Veteran" Jawa Timur.

	Text 18081010142_Cover.pdf Download (1MB)
	Text 18081010142_Bab 1.pdf Download (92kB)
	Text 18081010142_Bab 2.pdf Restricted to Repository staff only until 4 June 2027. Download (208kB)
	Text 18081010142_Bab 3.pdf Restricted to Repository staff only until 4 June 2027. Download (390kB)
	Text 18081010142_Bab 4.pdf Restricted to Repository staff only until 4 June 2027. Download (1MB)
	Text 18081010142_Bab 5.pdf Download (14kB)
	Text 18081010142_Dafpus.pdf Download (98kB)

Abstract

This study aims to develop a sugarcane leaf disease image classification system using a web-based transfer learning approach. The dataset consists of five classes of sugarcane leaf conditions (healthy, Red Rot, Red Stripe, Rust, and Bacterial Blight) obtained from the Kaggle platform. Five popular pretrained architectures were evaluated with two learning rate variations (0.0001 and 0.00001) using accuracy, precision, recall, and F1-score metrics. The results indicate that the DenseNet121 architecture with a learning rate of 0.00001 achieved the best accuracy on the test data. The best-performing model was integrated into a Flask based web application as a tool for early detection of sugarcane leaf diseases. This finding is expected to support efforts to improve sugarcane agricultural productivity through faster and more accurate disease diagnosis.

Item Type:

Thesis (Undergraduate)

Contributors: