Claim Missing Document
Check
Articles

Found 1 Documents
Search
Journal : Compiler

Baseline Evaluation of Backpropagation Artificial Neural Network for Visual Image-Based Vehicle Type Classification Harman, Rika; Riadi, Imam; Fadlil, Abdul
Compiler Vol 14, No 2 (2025)
Publisher : Institut Teknologi Dirgantara Adisutjipto

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.28989/compiler.v14i2.3210

Abstract

The increasing number of vehicles in urban areas requires technology-based solutions for efficient transportation management. This study proposes a vehicle classification model using Artificial Neural Networks (ANN) with the backpropagation algorithm, based on digital image data. The model is a feedforward neural network comprising an input layer, a hidden layer with 64 sigmoid-activated neurons, and an output layer with 7 softmax-activated neurons. The dataset, sourced from Roboflow Inc., consists of 16,185 images across eight vehicle classes: Hummer, Toyota Innova, Hyundai Creta, Suzuki Swift, Audi, Mahindra Scorpio, Rolls Royce, and Tata Safari. The data is split 80:20 for training and testing. Input features include vehicle dimensions, dominant RGB color, number of axles, and license plate detection. The model is trained using gradient descent and categorical crossentropy loss. Evaluation results show 85% validation accuracy at epoch 28 and 100% test accuracy. Precision, recall, and F1-score indicate strong performance, though minor errors occur in visually similar classes. These findings demonstrate that backpropagation-based ANN is effective for vehicle classification and can be applied in systems such as automatic parking and traffic monitoring