A Novel Word Spotting Algorithm Using Bidirectional Long Short-Term Memory Neural Networks

Keyword spotting refers to the process of retrieving all instances of a given key word in a document. In the present paper, a novel keyword spotting system for handwritten documents is described. It is derived from a neural network based system for unconstrained handwriting recognition. As such it performs template-free spotting, i.e. it is not necessary for a keyword to appear in the training set. The keyword spotting is done using a modification of the CTC Token Passing algorithm. We demonstrate that such a system has the potential for high performance. For example, a precision of 95% at 50% recall is reached for the 4,000 most frequent words on the IAM offline handwriting database.