The Bangladesh road traffic sign dataset in real-world images for traffic sign recognition

Traffic sign detection and classification have significant impacts in the field of automated driving system, traffic management, driver assistance system, to detect traffic rules violations etc. In this paper, we have presented the Bangladesh road traffic sign benchmark dataset, which consists of 10259 real-world traffic sign images captured from various locations in Bangladesh and 10259 annotated images. A Total of 31 distinct traffic sign images were collected including Crossroad, Emergency Stopping, Sharp left turn. For image annotation, a sophisticated tool, Roboflow, has been utilized and data augmentation techniques have been applied to enhance the diversity of the images. The dataset is useful for training and testing of any deep convolutional neural networks (CNNs) models for traffic sign recognition. The dataset is publicly accessible via the following link: https://zenodo.org/records/14969122.