Enhancing the Traffic Sign Detection to Improve Convolutional Neural Network

Traffic signs are essential for controlling traffic upon the road, enforcing driver discipline, and averting accidents, property damage, and deaths. An integral component of any ITS (Intelligent Transportation System) is road sign management with automated detection and recognition. The sign could h...

Full description

Saved in:

Bibliographic Details
Published in:	2024 Second International Conference on Advances in Information Technology (ICAIT) Vol. 1; pp. 1 - 6
Main Authors:	Sungheetha, Akey, V, Vijeya kaveri, V, Sruthi Sri, S, Srineveda R, K, Sasikal Rani, R, Rajesh Sharma, Antony, Judeson
Format:	Conference Proceeding
Language:	English
Published:	IEEE 24-07-2024
Subjects:	CNN Architecture Convolution Convolutional neural networks Deep Learning Feature extraction Information technology Precision Rendering (computer graphics) Road Safety Roads Robustness Safety Timing Traffic control Traffic Sign Detection Vehicles
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	Traffic signs are essential for controlling traffic upon the road, enforcing driver discipline, and averting accidents, property damage, and deaths. An integral component of any ITS (Intelligent Transportation System) is road sign management with automated detection and recognition. The sign could have a large variety of colors and sizes. Additionally, changes may result from the signs' translation, rotation, and scaling. Transportation officials can optimize traffic control by modifying ramp metering, adjusting signal timings, or providing congestion-aware route guidance based on these traffic flow predictions. The skip connections are created by using dilated convolutions with various filter sizes to broaden the feature extractors' receptive fields. It was suggested to make an early effort to create a hierarchical, spatially invariant model for image classification. A novel road extraction framework is proposed to address these issues by extracting complete and continuous road networks using a worldwide Context-aware and Batch-independent Network. The precision attained by the suggested model is superior to that of the conventional CNN. Furthermore, the suggested model demonstrates resilient performance in various environmental circumstances, rendering it appropriate for practical implementations and advancing intelligent transportation systems development, ultimately advocating for safer roadways.
DOI:	10.1109/ICAIT61638.2024.10690365