Accuracy Improvement of Automatic Smoky Diesel Vehicle Detection Using YOLO Model, Matching, and Refinement
View abstract on PubMed
Summary
This summary is machine-generated.A new method accurately detects smoky diesel vehicles using advanced object detection and image processing. This approach significantly improves detection rates and reduces processing time for cleaner transportation.
Area Of Science
- Environmental Science
- Computer Vision
- Transportation Engineering
Background
- Transportation is a major source of air pollution, with smoky diesel vehicles being a significant contributor.
- Effective detection of these vehicles is crucial for implementing targeted emission reduction strategies.
Purpose Of The Study
- To develop and evaluate a novel, highly accurate method for detecting smoky diesel vehicles.
- To improve upon existing methods in terms of accuracy, robustness, and processing speed.
Main Methods
- Utilized You Only Look Once (YOLO) models, specifically YOLOv5s, for initial object detection of vehicles, license plates, and smoke.
- Implemented two matching techniques correlating smoke regions with vehicle shapes and license plates.
- Applied image processing for refining smoke region detection and eliminating false positives caused by shadows.
Main Results
- YOLOv5s achieved high precision (91.4%) and mean average precision (mAP@0.5) of 91% for smoke region detection.
- The proposed method reached a detection rate of 97.45% and precision of 93.50%, outperforming existing methods.
- Processing time was reduced to 85 ms per image, significantly faster than reference studies.
Conclusions
- The developed method demonstrates superior accuracy, robustness, and efficiency in detecting smoky diesel vehicles.
- The system's performance indicates strong potential for real-world application in air pollution monitoring and control.
- This advancement contributes to reducing transportation-related air pollution through improved vehicle emission detection.

