Abstract
Speed limits, no entry, stoplights, turning left or right, minors crossing, heavy trucks not passing, and many more road signs exist. Objects such as trees, automobiles, people, bikes, and so on are spotted on the road. In this project, a deep neural network model is developed to identify traffic signs in photos using the GTSRB (German Traffic Sign Database) dataset, and items such as bikes, pedestrians, and vehicles are spotted using the COCO dataset. This research aims to train the CNN model to examine more traffic signs in potentially adverse settings such as bad weather and blurred images. To achieve maximum accuracy compared to existing models and to optimize anti-error recognition, the LeNet-5 CNN model is used for the GTRSB dataset, and Mask RCNN is used for the COCO dataset.