Prediksi Kerusakan Bodi Mobil Menggunakan CNN

Anutfaqiha, Afif Ramadlani (2024) Prediksi Kerusakan Bodi Mobil Menggunakan CNN. Undergraduate thesis, Universitas Muhammadiyah Malang.

[thumbnail of PENDAHULUAN.pdf]

Download (937kB) | Preview
[thumbnail of BAB 1.pdf]
BAB 1.pdf

Download (94kB) | Preview
[thumbnail of BAB 2.pdf]
BAB 2.pdf

Download (511kB) | Preview
[thumbnail of BAB 3.pdf]
BAB 3.pdf

Download (454kB) | Preview
[thumbnail of BAB 4.pdf] Text
BAB 4.pdf
Restricted to Registered users only

Download (1MB) | Request a copy
[thumbnail of BAB 5.pdf] Text
BAB 5.pdf
Restricted to Registered users only

Download (83kB) | Request a copy
[thumbnail of LAMPIRAN.pdf] Text
Restricted to Registered users only

Download (173kB) | Request a copy
[thumbnail of POSTER.pdf] Text
Restricted to Registered users only

Download (170kB) | Request a copy


Vehicle damage detection is one of the key activities that are important in the insurance and vehicle rental industries. Such systems are widely used to identify vehicle damage after an accident has occurred by the driver and also by insurance companies to detect and determine the appropriate evaluation of damage and vehicle rental companies to assign vehicle damage fees to at-fault customers. The core technique based on these systems is object recognition. However, object recognition and classification being a confusing research range, the reliability of this kind of project lies in the feature selection and extraction procedure.
The aim of this study is to create a system for classifying the level of damage in cars, utilizing CNN methodology to achieve accuracy in both results and object detection. This research employs a descriptive qualitative approach, employing observational data collection that involves scrutinizing objects by observing the prevailing situational conditions and events.
This research presents a highly accurate 2D image-based vehicle damage detection approach. In this system, the accuracy level is processed by artificial neural networks. The CNN method with a three-layer architecture, 3x3 kernel and a learning coefficient of 0.001 shows excellent performance, achieving a validation accuracy of 81.09% and a loss of 0.4634. Optimization by dividing the data into 70-20-10 and model parameters contributed to the success, with 3-layer CNN, max pooling and 3x3 kernel being the best choice.

Item Type: Thesis (Undergraduate)
Student ID: 201710130311001
Keywords: Analysis; Vehicle Damage; Vehicle; Digital Image; CNN;
Subjects: T Technology > T Technology (General)
Divisions: Faculty of Engineering > Department of Electrical Engineering (20201)
Depositing User: 201710130311001 afiframadlani
Date Deposited: 18 Jan 2024 01:11
Last Modified: 18 Jan 2024 01:11

Actions (login required)

View Item
View Item