Naufaldillah, Muhammad (2024) Pengukuran Kemiripan Gambar dengan ResNet50 dan Ekstraksi Alamat Dari Citra Streeview Untuk Identifikasi Lokasi Rumah. Other thesis, Institut Teknologi Sepuluh Nopember.
Text
05111940000202-Undergradiate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 July 2026. Download (4MB) | Request a copy |
Abstract
Permintaan air bersih di Surabaya sangat meningkat. Perusahaan Daerah Air Minum perlu pengelolaan aset yang optimal untuk mempertahankan tingkat layanan yang diinginkan.. Tetapi kendala yang dihadapi adalah data pelanggan PDAM belum memiliki gambar. Solusinya melakukan profiling pelanggan PDAM berdasarkan jenis bangunan menggunakan citra street view. Penggunaan citra street view semakin populer sebagai sumber informasi dalam bentuk gambar panorama. Tugas akhir ini mengajukan metode pengukuran kemiripan antar citra dapat dilakukan dengan algoritma cosine similarity dengan metode deep learning ResNet50. Kemudian dilakukan georeference dengan menggunakan Google Maps API untuk mendapatkan informasi alamat yang sesuai dengan koordinat yang ada pada data pelanggan pada gambar tersisa. Rencana uji coba akan dilakukan di daerah Sememi Jaya, Wonorejo Permai Timur, dan Tambak Medokan Ayu. Untuk mendapatkan rumah dari gambar diperlukan perbandingan 10 gambar ke depan dan 10 gambar ke belakang. Terdapat dua parameter yang harus diatur secara manual, yaitu kecepatan mobil saat mengambil gambar dan interval waktu pengambilan gambar. Hasil yang didapatkan berupa gambar citra street view beserta alamat yang didapatkan dari hasil ekstraksi alamat dan disimpan dalam bentuk .jpg. Untuk model klasifikasi jenis pelanggan PDAM dimana merupakan kelas rumah atau usaha dengan data foto dengan pembuatan dataset dengan Roboflow, didapatkan nilai akurasi mAP sebesar 73,3 %, presisi sebesar 57,4%, dan recall sebesar 84,2%. Untuk pengukuran kemiripan gambar menggunakan model algoritma ResNet50, dari 690 gambar dalam dataset PDAM Surya Sembada Kota Surabaya, tersisa 468 gambar yang dipertahankan untuk analisis lokasi menggunakan Google Maps API
===================================================================================================================================
The demand for clean water in Surabaya is increasing greatly. Regional Drinking Water Companies need optimal asset management to maintain the desired level of service. But the obstacle faced is that PDAM customer data does not yet have a picture. The solution is to profile PDAM customers based on building type using street view imagery. The use of street view imagery is increasingly popular as a source of information in the form of panoramic images. This final project proposes a method for measuring similarity between images that can be done using the cosine similarity algorithm with the ResNet50 deep learning method. Then georeferencing is carried out using the Google Maps API to obtain address information that matches the coordinates in the customer data in the remaining image. The trial plan will be carried out in the Sememi Jaya, Wonorejo Permai Timur and Tambak Medokan Ayu areas. To get a house from an image, you need to compare 10 forward images and 10 backward images. There are two parameters that must be set manually, namely the speed of the car when taking pictures and the time interval for taking pictures. The results obtained are in the form of street view images along with addresses obtained from the address extraction results and saved in .jpg form. For the PDAM customer type classification model which is home or business class with photo data by creating a dataset with Roboflow, the mAP accuracy value was 73.3%, precision was 57.4%, and recall was 84.2%. To measure image similarity using the ResNet50 algorithm model, of the 690 images in the PDAM Surya Sembada Surabaya City dataset, the remaining 468 images were retained for location analysis using the Google Maps API
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Kemiripan Gambar, Kemiripan Kosinus, Deep Learning, Convolutional Neural Network; ResNet50, Google Street View, Google Maps API Image Similarity, Cosine Similarity, Deep Learning, Convolutional Neural Network, ResNet50, Google Street View, Google Maps API |
Subjects: | T Technology > T Technology (General) T Technology > T Technology (General) > T174 Technological forecasting T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T58.62 Decision support systems |
Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
Depositing User: | Muhammad Naufaldillah |
Date Deposited: | 06 Feb 2024 01:55 |
Last Modified: | 06 Feb 2024 01:55 |
URI: | http://repository.its.ac.id/id/eprint/106195 |
Actions (login required)
View Item |