ItalianSigns is a dataset containing images of the road context captured by phone camera. It contains 362 labelled images: each image contains a road sign (speed limit only), of which the number associated with the speed limit, and the bounding box corresponding to the ROI of the sign were annotated.

To recognize the road signs, we applied preprocessing techniques clean the image, used HoughCircles to find the spatial location of the sign and extract the bounding box, and used a KNN on feature vectors extracted via SIFT between ground truth images, and the crop of the inference image