Year: 2025
Type: article
Abstract: The realm of vision-based vehicle detection is currently bifurcated into traditional machine vision approaches and intricate deep learning methodologies.Traditional methods rely on motion-based techniques to discern vehicles from a static background image, encompassing strategies such as background subtraction, continuous video frame difference, and optical flow analysis.While traditional machine vision methods exhibit faster processing speeds, they struggle with variations in lighting, periodic background motion, and complex scenes.Conversely, deep learning-based approaches offer enhanced accuracy and adaptability but may encounter challenges with scale variations and small object detection.Vehicle detection, a critical task in various domains including transportation, surveillance, and autonomous driving, has witnessed significant advancements through the utilization of both machine learning (ML) and deep learning (DL) techniques.ML algorithms such as Support Vector Machines (SVMs) and Random Forests have traditionally been employed for feature-based vehicle detection, offering robustness and interpretability.This paper projects challenges and future directions in vehicle detection using image processing and deep learning.
DOI: 10.70505/ijwer96
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4414353557
FWCI:
Cites:
Cited by:
Related to: