Authors :
Harikesh Kumar Sharma; Anupama Jamwal
Volume/Issue :
Volume 8 - 2023, Issue 11 - November
Google Scholar :
https://tinyurl.com/msk7tj5f
Scribd :
https://tinyurl.com/58383jt9
DOI :
https://doi.org/10.5281/zenodo.10159118
Abstract :
This paper shows the development and
working of a traffic sign recognition system using
Convolutional Neural Networks (CNNs) for the accurate
detection and classification of traffic signs from images
(captured by a camera mounted on a vehicle). The system
achieved a high classification accuracy of 95% on a
validation set and consists of three main components:
image pre-processing, CNN model, and post-processing.
The pre-processing step involves resizing, normalization,
and augmentation, while the CNN model uses multiple
convolutional and fully connected layers to classify input
images into different traffic sign categories. Post-
processing involves non-maximum suppression and
thresholding to remove duplicate detections and filter out
false positives. Testing on a real-world dataset resulted in
an overall accuracy of 94.5%, demonstrating the system's
robustness to varying lighting and weather conditions, as
well as its ability to accurately recognize traffic signs at
different distances and angles. This proposed system
using CNNs is a promising solution for improving road
safety and reducing accidents caused by driver error.
Keywords :
Traffic Sign Recognition, Convolutional Neural Network , Traffic Sign Detection, Deep Learning.
This paper shows the development and
working of a traffic sign recognition system using
Convolutional Neural Networks (CNNs) for the accurate
detection and classification of traffic signs from images
(captured by a camera mounted on a vehicle). The system
achieved a high classification accuracy of 95% on a
validation set and consists of three main components:
image pre-processing, CNN model, and post-processing.
The pre-processing step involves resizing, normalization,
and augmentation, while the CNN model uses multiple
convolutional and fully connected layers to classify input
images into different traffic sign categories. Post-
processing involves non-maximum suppression and
thresholding to remove duplicate detections and filter out
false positives. Testing on a real-world dataset resulted in
an overall accuracy of 94.5%, demonstrating the system's
robustness to varying lighting and weather conditions, as
well as its ability to accurately recognize traffic signs at
different distances and angles. This proposed system
using CNNs is a promising solution for improving road
safety and reducing accidents caused by driver error.
Keywords :
Traffic Sign Recognition, Convolutional Neural Network , Traffic Sign Detection, Deep Learning.
