Deteksi Objek Pada Gambar Dengan Intensitas Rendah Menggunakan Deep Learning

Mahendra, Rama Yusuf (2021) Deteksi Objek Pada Gambar Dengan Intensitas Rendah Menggunakan Deep Learning. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

Text 07211740000013-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2023. Download (26MB) | Request a copy

Abstract

Gambar intensitas rendah banyak ditemui pada kehidupan sehari-hari. Contoh paling banyak terdapat pada gambar yang diambil pada kondisi kurang cahaya. Kurangnya cahaya ini menyebabkan adanya kesulitan untuk mendeteksi suatu objek karena kurangnya visibilitas pada citra tersebut. Kontras antara objek dan latar belakang pun kurang tinggi sehingga ada kesulitan untuk membedakan keduanya. Untuk mengatasi masalah deteksi objek pada gambar intensitas rendah, diperlukan pendeteksi objek yang dirancang khusus untuk gambar intensitas rendah. Pendeteksi objek khusus gambar intensitas rendah ini dilatih menggunakan algoritma YOLO dengan Exclusively Dark Image Dataset yang berisikan 12 jenis objek dan 7.363 total gambar dataset. Dataset tersebut akan diproses dalam tiga tahap, yaitu training, validation, dan testing. Hasil dari tahapan tersebut berupa model yang diharapkan dapat mendeteksi objek pada gambar intensitas rendah dengan akurasi yang tinggi.
===================================================================================================
Low intensity image often to be found in daily life. The most example of it can be found in image that taken in a low light environment. This low light condition causes difficulties to detect some object because of lack of visibility in the image. Contrast between the object and background is also not high enough and causes difficulties to differentiate the two. To solve problem of object detection in low intensity image, a specially designed object detector for low intensity image is needed. The object detector for low intensity image is trained with YOLO algorithm and using Exclusively Dark Image Dataset that contain 12 object classes and total of 7.363 image dataset. The dataset will be processed in three steps, namely training, validation, and testing. The result of these steps is a model that we expected can detect object in low intensity image with high accuracy.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Intensitas Rendah, Kurang Cahaya, YOLO, Low Intensity, Low Light, YOLO
Subjects:	T Technology > T Technology (General) T Technology > TR Photography
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	Rama Yusuf Mahendra
Date Deposited:	01 Sep 2021 02:02
Last Modified:	01 Sep 2021 02:02
URI:	http://repository.its.ac.id/id/eprint/91209

Actions (login required)

View Item