Image of DETEKSI JALAN DI SEKITAR AUTONOMOUS ELECTRIC VEHICLE BERBASIS ALGORITMA CONVOLUTIONAL NEURAL NETWORK (CNN)

Skripsi

DETEKSI JALAN DI SEKITAR AUTONOMOUS ELECTRIC VEHICLE BERBASIS ALGORITMA CONVOLUTIONAL NEURAL NETWORK (CNN)

Penilaian

0,0

dari 5
Penilaian anda saat ini :  

An autonomous electric vehicle needs to know the road conditions while operating. The autonomous electric vehicle has to detect the lanes in order to run in the desired routes. Such lanes could be structured and non-structured which may affect affect the operation of the electric autonomous vehicle. This study addressed such problems using image processing and lanenet method based on Convolutional Neural Network (CNN). This combination can detect roads in real-time. In this study, we used two optimizers for training, namely Adam and stochatic gradient descent (SGD) to detect roads and non-roads, such as grass, soil, and sidewalks. The data used in this study were the roads around the faculty of engineering, Universitas Sriwijaya. The result showed that model 1 which used optimizer Adam, 200 epoch, and learning rate 0.001 gave the final loss value of 2.79. The test results showed that model 1 is better than model 2 which used SGD with an accuracy of 81.49%. Test was also carried for different images data with poor lighting and the accuracy was only 48.52%. The low accuracy was due to the image capture process using a lower resolution, framer to second, and an unstable camera. This study also compared the road detection using CNN and HSV. CNN gave an accuracy of 79.46% which is lower than HSV. However, CNN could detect the lane directly without converting the images into binary images as in HSV.

Availability
Inventory Code Barcode Call Number Location Status
2007001436T40360T403602020Central Library (REFERENCES)Available but not for loan - Not for Loan
Detail Information
Series Title
-
Call Number
T403602020
Publisher
Inderalaya : Prodi Teknik Elektro, Fakultas Teknik Universitas Sriwijaya.,
Collation
xvii, 75 hlm. : ilus. ; 28 cm
Language
Indonesia
ISBN/ISSN
-
Classification
625.707
Content Type
Text
Media Type
unmediated
Carrier Type
-
Edition
-
Subject(s)
teknik jalan raya
Jurusan Teknik Elektro
Specific Detail Info
-
Statement of Responsibility
Other version/related

No other version available

File Attachment
Comments
You must be logged in to post a comment