Paper Title: Vision-Based Black Ice Detection using Convolutional Neural Networks for Road Safety
Conference Name: ICMIC 2024
Abstract: Black ice poses a severe threat to road safety, particularly in cold climates, often leading to accidents and fatalities. Traditional detection methods lack accuracy and efficiency. This abstract introduces a vision-based black ice detection system employing deep learning. The system utilizes a convolutional neural network (CNN) trained on a diverse dataset of onboard camera images, encompassing various environmental conditions and road textures. Through data augmentation, transfer learning, and fine-tuning, the CNN learns to accurately identify black ice regions. Key features include real-time processing and integration with vehicle telematics for proactive warnings. Upon detection, the system alerts drivers via dashboard displays or smartphone apps, enabling timely response. Comprehensive testing demonstrates high accuracy and minimal false positives under diverse conditions. The system's adaptability to different vehicles and scalability across fleets enhances its practicality in mitigating black ice risks. the vision-based black ice detection system offers a proactive solution for identifying and alerting drivers to hazardous conditions, reducing weather-related accidents and enhancing transportation safety.