IoT-Enhanced Smart Parking Management With IncepDenseMobileNet for Improved Classification
With the fast progression of urbanization, efficient parking management is crucial for high-traffic areas like Los Angeles International Airport (LAX). This research introduces an AI-driven approach to improve smart parking using deep learning and IIoT-based monitoring for real-time analysis. The da...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/11080429/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | With the fast progression of urbanization, efficient parking management is crucial for high-traffic areas like Los Angeles International Airport (LAX). This research introduces an AI-driven approach to improve smart parking using deep learning and IIoT-based monitoring for real-time analysis. The data from January 2021 to June 2024 includes parking occupancy, vehicle classification, payment behaviors, and violations. Advanced preprocessing techniques, such as Temporal Variability Adjustment and Harmonic Noise Compensation, enhanced data quality, while Proportional Adaptive Balancing and Augmentation (PABA) addressed class imbalance. The Hybrid Adaptive Feature Selector (HAFS) enhanced critical attributes via statistical and genetic diversity techniques. The IncepDenseMobileNet (IDMN) model integrates Inception, DenseNet, and MobileNet architectures to proficiently capture complex patterns via multi-scale feature extraction and efficient depthwise separable convolutions. Simulations show that IDMNachieved remarkable results, including 98.6% accuracy, 0.95 precision, and 0.96 recall. The model demonstrated a log loss of 0.18 and outstanding performance on new metrics, including Weighted Error Impact Score (WEIS), Dynamic Class Stability Index (DCSI), and Harmonized Classification Risk (HCR), signifying its effectiveness. The examination of the confusion matrix indicated a high predictive accuracy, with AUC values over 97% for both “Occupancy Status” and “Vehicle Type.” Statistical testing validated the robustness of the technique, while sensitivity analysis identified optimal hyperparameter combinations. The findings indicate significant improvement in parking management via accurate prediction of occupancy trends, identification of peak demand, and detection of payment and violation issues. This paper presents an innovative framework for improving parking operations, dynamic pricing, and resource allocation in urban environments. |
|---|---|
| ISSN: | 2169-3536 |