Mathematical Optimization Group & PLANET intelligent systems GmbH

Computational Intelligence Technology Laboratory

Machine Learning Algorithms for Document and Text Recognition

ImageCLEF 2016

First Rank at ImageCLEF 2016 Handwritten Scanned Document Retrieval Task : Multi word query keyword search competition

Marvelous news: HORIZON-2020 project READ confirmed

According to today's confirmation, we are now looking forward to finally setting up the HORIZON-2020 project READ (Recognition & Enrichment of Archival Documents) ... with CITlab-PLANET's essential ARGUS software for text recognition in historical documents!

Two new projects confirmed

In Spring 2014, CITlab received two new project fundings for our ongoing work in CI-based recognition technology.

Rank 3 at OpenHaRT 2013 Contest

We achieved Rank 3 in the constrained DIR task of the 2013 NIST Open Handwriting Recognition and Translation Evaluation (with ICDAR 2013), a document recognition evaluation for handwritten transcriptions of arabic news.

Official Evaluation Results

Pole position again: Two first ranks at ICDAR 2015 competitions

At ICDAR 2015 – this year's leading document analysis & recognition event – the CITlab-PLANET project team participated in two competitions ... and again ... very successful!

  • First Rank at HTRtS-2015: Handwritten Text Recognition on the tranScriptorium Dataset (Restricted Track)
  • First Rank at KWS-2015: Keyword Spotting for Handwritten Documents (Training-based Track)

 

 

Five first ranks at the ANWRESH-2014 Contest

The CITlab–PLANET project team received 1. ranks in all of the five ANWRESH 2014 (with ICFHR 2014) evaluations on Word Recognition from Segmented Historical Documents.

Official Evaluation Results

Rank 1 at the HTRtS-2014 Contest

We achieved the First Rank in the restricted scheme of the 2014 HTRtS contest (with ICFHR 2014), a document recognition evaluation for Handwriting Text Recognition in a transScriptorium Dataset.

 Official Evaluation Results (see Results section)