Abstract:
In this paper, we propose a novel methodology to classify traffic signs with the purpose of boosting the classification accuracy. Our model consists of two parts: one is ...Show MoreMetadata
Abstract:
In this paper, we propose a novel methodology to classify traffic signs with the purpose of boosting the classification accuracy. Our model consists of two parts: one is image data preprocessing, the other one is a modified Residual Networks (mResNets). The image data preprocessing includes color space conversion, data augmentation, and data normalization. The modified Residual Networks yields a competitive performance. The experimental result shows the robustness of our model and its superiority. We have achieved the excellent performance of 99.66% on the German traffic sign recognition Benchmark (GTSRB) dataset.
Date of Conference: 11-14 December 2017
Date Added to IEEE Xplore: 05 February 2018
ISBN Information:
Electronic ISSN: 2474-2325