Download

Publications, presentations, and patents using this database must cite the papers:
Fredrik Larsson and Michael Felsberg , Using Fourier Descriptors and Spatial Models for Traffic Sign Recognition , In Proceedings of the 17th Scandinavian Conference on Image Analysis, SCIA 2011, LNCS 6688, pp. 238-249. bib , doi:10.1007/978-3-642-21227-7_23
Fredrik Larsson, Michael Felsberg, and Per-Erik Forssen , Correlating Fourier descriptors of local patches for road sign recognition , IET Computer Vision, 5(4): 244-254, 2011. bib , doi:10.1049/iet-cvi.2010.0040
Without prior approval from CVL, the Traffic Signs Database, in whole or in part, may not be modified or used for commercial purposes.
A reported match is to be considered true if the overlap error (1-intersection/union) between reported and annotated bounding box is less than 0.5. Parameter tuning/machine learning is to be done on the set that is not used for evaluation.
Let us know if you get any results you would like us to add to the precision-recall table.

We currently provide Matlab code for parsing the annotation files and displaying the results. Part0 for each set contains the annotated images (roughly 2000 images in a 1.5Gb zip file) while Part1-Part4 contains the not annotated images.

Instructions to get started.

1. Download code.zip

2. Download Part0.zip and annotations.txt for one of the datasets. Extract the images and put the annotation file in the same directory

3. Change the base path in example.m so that it points to your image folder

4. Run example.m (the two different sets need to be kept in different folders, e.g. /Set1 and /Set2, for example.m to work).