Deep Learning Framework for Automated Vehicle Classification and Counting Using Road Energy-Harvesting Sensors Data
Traffic congestion and ineffective vehicle monitoring systems remain pressing challenges in urban transportation which is leading to time delays, accidents, and reduced road efficiency. Conventional approaches such as manual counting, inductive loop sensors, and traditional image processing often la...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English English |
| Published: |
https://www.jatit.org
2025
|
| Subjects: | |
| Online Access: | http://ir.unimas.my/id/eprint/49957/ https://www.jatit.org/index.php |
| Abstract | Abstract here |
| Summary: | Traffic congestion and ineffective vehicle monitoring systems remain pressing challenges in urban transportation which is leading to time delays, accidents, and reduced road efficiency. Conventional approaches such as manual counting, inductive loop sensors, and traditional image processing often lack accuracy, scalability, and adaptability to real-world conditions. These limitations create a research gap for automated, reliable, and intelligent solutions. This study proposes a deep learning based framework for automated vehicle classification and counting using convolutional neural networks (CNNs). Using piezoelectric sensors, a numerical data of road traffic is collected, pre-processed, and augmented to enhance data quality and reduce overfitting. The 1D-CNN model is trained to classify multiple vehicle categories and integrated into a counting mechanism. Model performance is evaluated using standard metrics, including accuracy, precision, recall, F1-score. Achieving high classification accuracy across diverse traffic conditions, these experimental results show that the proposed framework significantly outperforms conventional machine learning and image processing methods. This research contributes both academically and practically by providing: (1) a CNN-based methodology tailored for vehicle detection and counting, (2) a benchmark dataset for evaluation, and (3) an effective framework that can support intelligent transportation systems (ITS). |
|---|