TraffSign: Multilingual Traffic Signboard Text Detection and Recognition for Urdu and English

Scene-text detection and recognition methods have demonstrated remarkable performance on standard benchmark datasets. These methods can be utilized in human-driven/self-driving cars to perform navigation assistance through traffic signboard text detection and recognition. Existing datasets include scripts in numerous languages like English, Chinese, French, Arabic, German, etc. However, traffic navigation signboards in Pakistan and many states of India are written in Urdu along with the English translation to guide human drivers. To this end, we present Deep Learning Laboratory’s Traffic Signboards Dataset (DLL-TraffSiD) to develop multi-lingual text detection and recognition methods for traffic signboards. In addition, we present a pipeline for multi-lingual text detection and recognition for an outdoor road environment. The results show that our presented system signified better applicability in text-detection and text recognition, and achieved 89% and 92.18% accuracy on the proposed dataset (The proposed dataset along with implementation is available at https://github.com/aatiibutt/TraffSign/).