Multi-column deep neural network for traffic sign classification
At a glance
Combined neural networks to recognize traffic signs, exceeding the benchmark's human comparison score.
In one paragraph
Combined several neural networks trained on differently processed images to recognize traffic signs. The system won the German Traffic Sign Recognition Benchmark with 99.46% accuracy, surpassing the human comparison score on that test and demonstrating the potential of GPU-trained deep learning.
Authors 4
- Dan Ciresan
- Ueli Meier
- Jonathan Masci
- Jürgen Schmidhuber
