Interactive-Predictive Handwritten Text Recognition

Interactive-Predictive Handwritten Text Recognition (HTR) of historical documents
is showed here. In the Interactive-Predictive HTR approach, the user and the system
work in tight mutual collaboration to obtain the transcription of the given image:
each time the user amends an error, the system proposes a new hypothesis taken into
account the current prefix (implicitly) validated by the user for proposing a better
suffix. Several prototypes are available related with three collections. In all prototypes
the images were not included for training HTR models.