Pengaruh Algoritma Deep Learning dalam Meningkatkan Akurasi Sistem Pendeteksian Kondisi Jalan Raya

Sari

The condition of the public road surface is one of the key factors in traffic safety and security. Lack of maintenance and heavy traffic condition will affect to the road surface. The process of preventing damage is very important to minimize the bad effect the road. Nowadays, the development of deep learning and high computing capabilities make it possible to automate the process of detecting an object. In this study, we carry out the process of detecting and recognizing road conditions using an artificial neural network combined with deep learning to maximize prediction results. The results of this model can be used continuously for the process of surveying and mapping road conditions. In this work, we extract the input images and then study them through an artificial neural network system. The automatic learning process will be run with a deep learning method using the Convolutional Neural Network (CNN) based on VGG network. Through this approach, this model can detect and recognize differences in road conditions very well. The effectiveness of this model is applied to the original image data that has been collected by the Pothole dataset with a prediction accuracy rate of 97.3% at comparison level 80% and 20% feature extraction during training, validation and testing respectively.