A Novel Approach to Road Safety: Detecting Illegal Overtaking Using Smartphone Cameras and Deep Learning for Vehicle Auditing

Journal of Sensor and Actuator Networks, Journal Year: 2025, Volume and Issue: 14(1), P. 10 - 10

Published: Jan. 26, 2025

Overtaking relies heavily on the driver’s attention and cognitive state, illegal overtaking can lead to accidents, severe injuries, or fatalities. To enhance highway safety, we propose a method for accurately detecting continuous road lanes. We used dashboard-mounted smartphone cameras geolocation data filter analysis areas. state-of-the-art deep learning model You Only Look Once version 8 (YOLOv8) detect yellow When these lanes suggest potential overtaking, apply YOLO Panoptic driving Perception 2 (YOLOPv2) model, followed by post-processing. confirm events checking overlaps between detections from both models. store confirmed instances evaluate information temporally rather than just individual frames. then analyze entire video identify violations extract moments of occurrence. tested algorithm real-world traffic under various weather lighting conditions. Our demonstrates reliability consistency in identifying overtaking. achieved 16 TP only 1 FP over 56 videos totaling 41 h, 9 min, 24 s, with precision, recall, F1-score values 1.000, 0.941, 0.970, respectively. Consequently, our innovative practical solution, utilizing simple advanced computer vision models, significantly safety support vehicle auditing systems.

Language: Английский

Effective lane detection on complex roads with convolutional attention mechanism in autonomous vehicles

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Aug. 19, 2024

Autonomous Vehicles (AV's) have achieved more popularity in vehicular technology recent years. For the development of secure and safe driving, these AV's help to reduce uncertainties such as crashes, heavy traffic, pedestrian behaviours, random objects, lane detection, different types roads their surrounding environments. In AV's, Lane Detection is one most important aspects which helps holding guidance departure warning. From Literature, it observed that existing deep learning models perform better on well maintained favourable weather conditions. However, performance extreme conditions curvy need focus. The proposed work focuses presenting an accurate detection approach poor roads, particularly those with curves, broken lanes, or no markings Convolutional Attention Mechanism (LD-CAM) model achieve this outcome. method comprises encoder, enhanced convolution block attention module (E-CBAM), a decoder. encoder unit extracts input image features, while E-CBAM quality feature maps images extracted from decoder provides output without loss any information original image. carried out using distinct data three datasets called Tusimple for condition images, Curve Lanes curve lanes Cracks Potholes damaged road images. trained showcased improved attaining Accuracy 97.90%, Precision 98.92%, F1-Score IoU 98.50% Dice Co-efficient 98.80% both structured defective

Language: Английский