PENERAPAN ALGORITMA CONVOLUTIONAL NEURAL NETWORK (CNN) DALAM MENGIDENTIFIKASI LAHAN KOSONG KABUPATEN KULON PROGO BERBASIS CITRA GOOGLE EARTH

Zaini Miftah Prasetyahadi, NIM.: 18106050044 (2023) PENERAPAN ALGORITMA CONVOLUTIONAL NEURAL NETWORK (CNN) DALAM MENGIDENTIFIKASI LAHAN KOSONG KABUPATEN KULON PROGO BERBASIS CITRA GOOGLE EARTH. Skripsi thesis, UIN SUNAN KALIJAGA YOGYAKARTA.

Preview	Text (PENERAPAN ALGORITMA CONVOLUTIONAL NEURAL NETWORK (CNN) DALAM MENGIDENTIFIKASI LAHAN KOSONG KABUPATEN KULON PROGO BERBASIS CITRA GOOGLE EARTH) 18106050044_BAB-I_IV-atau-V_DAFTAR-PUSTAKA.pdf - Published Version Download (2MB) | Preview
	Text (PENERAPAN ALGORITMA CONVOLUTIONAL NEURAL NETWORK (CNN) DALAM MENGIDENTIFIKASI LAHAN KOSONG KABUPATEN KULON PROGO BERBASIS CITRA GOOGLE EARTH) 18106050044_BAB-II_sampai_SEBELUM-BAB-TERAKHIR.pdf - Published Version Restricted to Registered users only Download (3MB) | Request a copy

Abstract

Kulon Progo faces several challenges related to its land management. There have been significant changes in land use over time, including increased development in previously open areas or even on land that was previously considered neglected. These changes put pressure on land availability and create a need for more detailed and effective land management. In this context, land classification in Kulon Progo is important because it provides a deeper understanding of land changes and helps in planning more efficient and sustainable land use. In this study, the Convolutional Neural Network method was used to classify land in Kulon Progo, which consists of open empty land, open nonempty land, and closed land. The dataset was created using images taken from Google Earth, totaling 5480 images cropped to a size of 50 x 50 pixels. Before the model was trained, the dataset was processed through preprocessing stages including cropping, using RGB images, and then augmentation. The algorithm was then trained with 500 epochs, learning rates of 0.001, 0.0001, and 0.00001, and Adam and RMSprop optimizers. From the experimentation that has been done with certain parameters, the highest training accuracy obtained was 0.9807, produced with parameters: learning rate 0.0001, epoch 500, and optimizer Adam. Meanwhile, the highest validation accuracy obtained was 0.8608, produced with parameters: learning rate 0.0001, epoch 500, and optimizer RMSprop.

Share this knowledge with your friends :

Actions (login required)

View Item