Abstract:
Autonomous vehicle technology requires reliable traffic sign detection and recognition (TSDR). The importance of this task has led to extensive research and many promising methods in the literature. Most of these methods have been evaluated on clean and challenge-free datasets, overlooking the performance deterioration caused by different challenging conditions (CCs) that obscure traffic-sign images captured in the wild. This paper examines the TSDR problem under CCs and its performance degradation. We propose a CNN-based prior enhancement-focused TSDR framework. Our modular approach includes a CNN-based challenge classifier, Enhance-Net—an encoder-decoder CNN architecture for image enhancement—and two CNN architectures for sign-detection and classification. We propose a new training pipeline for Enhance-Net that enhances traffic sign regions instead of the whole image in challenging images subject to their accurate detection. Our approach was tested using CURE-TSD dataset of traffic videos from different CCs. Our method achieves 91.1% precision and 70.71% recall, 7.58% and 35.90% better than the benchmark. Our approach outperforms CNN-based TSDR methods by a large margin.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
