Convolutional Neural Networks Applied to House Numbers Digit Classification
This work addresses the problem of digit recognition in street view images for applications like automated address reading, but it is incremental as it builds on existing ConvNet architectures.
The paper tackled digit classification in real-world house numbers using convolutional neural networks (ConvNets), achieving a new state-of-the-art accuracy of 94.85% on the SVHN dataset, which represents a 45.2% error improvement.
We classify digits of real-world house numbers using convolutional neural networks (ConvNets). ConvNets are hierarchical feature learning neural networks whose structure is biologically inspired. Unlike many popular vision approaches that are hand-designed, ConvNets can automatically learn a unique set of features optimized for a given task. We augmented the traditional ConvNet architecture by learning multi-stage features and by using Lp pooling and establish a new state-of-the-art of 94.85% accuracy on the SVHN dataset (45.2% error improvement). Furthermore, we analyze the benefits of different pooling methods and multi-stage features in ConvNets. The source code and a tutorial are available at eblearn.sf.net.