Traffic Sign Recognition Using YOLO And RCNN
B Prasad Yadav, Jagat , Jahnavi Sivani Motapotula, Teja Sai Koka
The “Traffic Sign Recognition with YOLO-RCNN” system, investigates a novel approach to traffic
sign recognition for advanced driving systems by combining the speed and efficiency of the You Only
Look Once (YOLO) object detection framework with the precision and accuracy of Region-based
Convolutional Neural Networks (R-CNN). YOLO excels at rapidly identifying multiple traffic signs
within an image, providing an initial set of detections. However, its initial bounding box predictions
can sometimes be imprecise. To address this limitation, R-CNN is integrated as a post-processing
step. R-CNN refines the initial detections by adjusting the bounding box coordinates and significantly
improving the accuracy of traffic sign classification. This hybrid approach leverages the strengths of
both architectures, resulting in a system that achieves a superior balance between speed and accuracy.
The system's performance was rigorously evaluated on a diverse dataset encompassing a wide range of
traffic sign types and challenging environmental conditions, such as varying lighting and occlusions.
Experimental results demonstrate a significant improvement in both detection accuracy and processing
speed compared to traditional methods. This innovative approach not only enhances the performance
of traffic sign recognition systems but also paves the way for more robust and reliable solutions for
autonomous driving applications. Future research directions include: optimizing the model architecture
through techniques such as network pruning and quantization to further improve inference speed and
computational efficiency, expanding the training dataset to include more diverse and challenging
scenarios, such as adverse weather conditions and complex urban environments, to enhance the model's
generalization capabilities, and exploring the integration of advanced deep learning techniques, such as
attention mechanisms and transformer architectures, to further improve the model's ability to focus on
salient features and enhance its overall performance.
