The paper Multi-column Deep Neural Networks for Image Classification (pages 7-8) shows an attempt at recognizing the traffic signs, with lower error rates, by using multi-column deep neural networks.
Are Google cars using similar techniques of predicting signs using DNN, or are they using some other method?
I'm not sure what Google is using to perform that task, but most companies use region based convolutional neural nets to locate traffic signs and other objects.
But other companies use a deep neural network + bag of words approach to find objects.
See: Bag-of-Words Based Deep Neural Network for Image Retrieval which shows a general approach, to get the exact location you can use Feature Matching or Random Boxes.
