07 July 2017

Handwritten Text Recognition

Encouraging news from the European Recognition and Enrichment of Archival Documents project about a technology we've long been waiting for.

A Swedish project achieved "an average Character Error Rate (CER) of 7.0%.  When a dictionary is integrated into the recognition process, the CER can be as low as 5.5%."

Remember that's the character error rate. For a five letter word at 7% CER the percent of words correct would be 70%. Read Trolls and water spirits – transcribing Swedish folklore records with Handwritten Text Recognition at http://read.transkribus.eu/2017/06/30/transcribing-swedish-folklore-records-with-htr/.

A second example without error statistics is at http://read.transkribus.eu/2017/07/06/keyword-searching-in-handwritten-text-new-breakthrough/.

No comments: