Triantasari, Triantasari (2025) Klasifikasi Kebakaran Hutan dan Lahan dari Citra Remote sensing Menggunakan Algoritma Convolutional Neural Network (CNN). Other thesis, Universitas Islam Riau.

Text skripsi_203510084_watermark.pdf - Submitted Version Restricted to Registered users only Download (8MB) | Request a copy

Abstract

Forest and land fires are an environmental disaster that requires an accurate early detection system. This research compares three deep learning models CNN Custom, MobileNet, and LSTM for classification of forest fire images based on satellite images. All models were developed using TensorFlow and optimized using the Random Search method. The results showed that CNN Custom improved its accuracy from 96.73% to 98.51%, MobileNet from 94.94% to 95.24%, and LSTM from 92.52% to 94.49%. Evaluation was done using confusion matrix, classification report, and learning curve graph. Custom proved to be the most superior in classification accuracy and consistency. This research shows that deep learning models are effective for detecting fires from images and have the potential to be applied in real-time web-based detection systems.

Actions (login required)

View Item