Segmentasi Jalan Berbasis Superpixel Pada Data Video Dengan Metode Faster Regional – Convolutional Neural Network

Alfiyan, Allif (2021) Segmentasi Jalan Berbasis Superpixel Pada Data Video Dengan Metode Faster Regional – Convolutional Neural Network. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
06111740000097-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (1MB) | Request a copy
[img] Text
06111740000097-Undergraduate_Thesis.pdf
Restricted to Repository staff only until 1 October 2023.

Download (1MB) | Request a copy

Abstract

Segmentasi citra pada umumnya merupakan proses pengolahan citra yang mempartisi citra menjadi beberapa wilayah (region) atau objek. Segmentasi citra dapat dimanfaatkan dalam berbagai bidang pekerjaan, salah satu bentuk dalam penerapan segmentasi citra yaitu segmentasi jalan, dimana proses segmentasi jalan dapat mendorong perkembangan teknologi kendali otomatis. Dengan segmentasi jalan, kendaraan mampu mengenali mana jalan yang dapat dilalui kendaraan, dan mana yang tidak. Salah satu bentuk segmentasi jalan yang dapat digunakan adalah segmentasi jalan berbasis superpixel dengan dukungan metode Faster Regional – Convolutional Neural Network (Faster R – CNN) pada Deep Learning. Metode Faster R – CNN sendiri merupakan pengembangan algoritma klasifikasi yang menggabungkan Fast Regional – Convolutional Neural Network (Fast R-CNN) dengan Region Proposal Network (RPN). Alur segmentasi jalan pada penelitian ini adalah input data video, akuisi data video, segmentasi superpixel, segmentasi jalan dengan metode Faster R – CNN dan penyatuan kembali frame menjadi data video. Pada tahap training, penelitian ini menggunakan data citra jalan dari KITTI dataset yang terdiri dari 289 citra. Sedangkan pada tahap testing menggunakan tiga buah video dokumentasi perjalanan yang diambil oleh penulis secara manual. Berdasarkan tahap uji coba yang dilakukan, didapatkan rata-rata nilai Accuracy sebesar 95,15% dan rata-rata nilai F-measure sebesar 93,59%. ================================================================================================ Image segmentation is generally an image processing process that partitions an image into several regions or objects. Image segmentation can be used in various fields of work, one of the forms in the application of image segmentation is road segmentation, where the road segmentation process can encourage the development of automatic control technology. With road segmentation, vehicles are able to recognize which roads can be passed by vehicles, and which are not. One form of road segmentation that can be used is superpixel-based road segmentation with the support of the Faster Regional – Convolutional Neural Network (Faster R – CNN) method in Deep Learning. The Faster R – CNN method itself is a classification algorithm development that combines the Fast Regional – Convolutional Neural Network (Fast R-CNN) with the Region Proposal Network (RPN). The road segmentation flow in this study is video data input, video data acquisition, superpixel segmentation, road segmentation with Faster R - CNN method and reunification of frames into video data. At the training stage, this research uses road image data from the KITTI dataset consisting of 289 images. Meanwhile, at the testing stage, three travel documentation videos were taken by the author manually. Based on the trial phase, the average Accuracy value was 95.15% and the F-measure average was 93.59%.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: road segmentation, Superpixel – Based Segmentation, Faster Regional – Convolutional Neural Network, Deep Learning, segmentasi jalan
Subjects: Q Science > QA Mathematics
Q Science > QA Mathematics > QA336 Artificial Intelligence
Q Science > QA Mathematics > QA76.6 Computer programming.
Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions: Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis
Depositing User: Allif Alfiyan
Date Deposited: 28 Aug 2021 17:59
Last Modified: 28 Aug 2021 17:59
URI: https://repository.its.ac.id/id/eprint/90254

Actions (login required)

View Item View Item