InkCapture is a tool designed specifically for handwriting recognition. It will offer not only the extraction of text from a document, but also advanced search in documents, where it will not only search for an exact match, but also similar words. This will allow you to significantly reduce the number of pages that need to be studied when searching for a particular term. In the searched subset of pages, it evaluates the degree of certainty of finding the searched text and thus gives the researcher the most precise and detailed information about where to start studying, and where the greatest probability of success lies. It also allows you to search by image, which searches the document for text that is visually similar to the specified image. The key innovation of InkCapture is its continuous learning and improvement of skills based on feedback from text recognition and search.
InkCapture system features
- Automatic detection and marking of areas within handwritten text
- Automatic recognition of handwritten text in block letters
- Automatic recognition of text written by hand
- Advanced document search based on similarity of terms – not only exact matches are searched, but also words similar to the search term
- Search not only based on the entered text, but also based on the image (the document searches for text visually similar to the entered image)
- Specialized focus on the Czech language – recognition and search with regard to diacritics and the Czech dictionary
If you would like to help shape the future of digitization, you can also participate directly in this project by providing your historical documents for InkCapture development. We will professionally digitize the documents for you, and utilize the data from them for training neural networks, which will be the basis for this new-generation technology.