Abstract

The recognition of traffic signs in many state-of-the-art driver assistance systems is performed by statistical pattern classification methods. Traffic signs in European countries share many similarities but also vary in colour, size, and depicted symbols, making it hard to obtain one general classifier with good performance in all countries. Training separate classifiers for each country requires huge amounts of labelled training data. A well-trained classifier for one country can be adapted to other countries by semi-supervised learning methods to perform reasonably well with relatively low requirements regarding labelled training data. Self-training classifiers adapting themselves to unknown domains always risk that the adaption will become ineffective or even fail completely due to the occurrence of incorrectly labelled samples. To assure that self-training classifiers adapt themself correctly, advanced multi-classifier training methods like co-training are applied. Document type: Part of book or chapter of book