Visual object recognition is often conceived of as a final step in a visual processing system, First, physical information in the raw image is used to isolate and enhance to-be-recognized clumps and then each of the resulting preprocessed representations is fed into the recognizer. This general conception fails when there are no reliable physical cues for isolating the objects, such as when objects overlap. This paper describes an approach, called centered object integrated segmentation and recognition (COISR), for integrating object segmentation and recognition within a single neural network. The application is handprinted character recognition. The approach uses a backpropagation network that scans a field of characters and is trained to recognize whether it is centered over a single character or between characters. When it is centered over a character, the net classifies the character. The approach is tested on a dataset of handprinted digits and high accuracy rates are reported.

We report on results of training backpropagation nets with samples of hand-printed digits scanned off of bank checks and hand-printed letters interactively entered into a computer through a stylus digitizer. Generalization results are reported as a function of training set size and network capacity. Given a large training set, and a net with sufficient capacity to achieve high performance on the training set, nets typically achieved error rates of 4-5% at a 0% reject rate and 1-2% at a 10% reject rate. The topology and capacity of the system, as measured by the number of connections in the net, have surprisingly little effect on generalization. For those developing hand-printed character recognition systems, these results suggest that a large and representative training sample may be the single, most important factor in achieving high recognition accuracy. Benefits of reducing the number of net connections, other than improving generalization, are discussed.