A satellite workshop to be held in conjunction with
ICDAR 2017
16th International Conference on Document Analysis and Recognition
Kyoto, Japan
Saturday, November 11, 2017
1100 - 1120
Deep Convolutional
Recurrent Network for Segmentation-free Offline Handwritten Japanese Text
Recognition
Nam Tuan Ly,Cuong Tuan Nguyen, Cong Kha Nguyen and
Masaki Nakagawa
1120 - 1140
DeepKHATT: A Deep Learning Benchmark on Arabic
Script
Riaz Ahmad, Saeeda Naz,
M. Zeshan Afzal, S. Faisal Rashid, Marcus Liwicki and Andreas Dengel
1140 - 1200
Detection and Recognition of Arabic Text in Video Frames
Wataru Ohyama, Seiya
Iwata, Tetsushi Wakabayashi and Fumitaka Kimura
1200 - 1220
The Impact of Visual Similarities of Arabic-like scripts
in Terms of Learning in an OCR System
Riaz Ahmad, Saeeda Naz,
M. Zeeshan Afzal, Shiekh Faisal Rashid, Markus Liwicki and Andreas Dengel
1230 - 1430
Lunch
Oral Session 2
1430 - 1450
Implicit Language Model in LSTM for OCR
Ekraam Sabir, Stephen
Rawls and Prem Natarajan
1450 - 1510
An Empirical Study of Effectiveness of Post-processing in
Indic Scripts
V S Vinitha, C V
Jawahar and Minesh Mathew
1510 - 1530
Improving Classical OCRs for Brahmic Scripts using Script
Grammar Learning
Dipankar
Ganguly, Sumeet Agarwal and Santanu Chaudhury
1530 - 1550
Benchmarking Scene Text Recognition in Devanagari, Telugu
and Malayalam
Minesh Mathew, Mohit
Jain and C V Jawahar
1550 - 1620
Coffee Break
1620 - 1630
Closing Remarks
The workshop will provide a forum for highlighting current research on multilingual document analysis systems with particular emphasis on OCR. The predecessors to this workshop were held in conjunction with ICDAR 1999 in Bangalore, India, ICDAR 2009 in Barcelona, Spain, ICDAR 2011 in Beijing, China, ICDAR 2013 in Washington DC, USA and ICDAR 2015 in Nancy, France. A joint Workshop on Multilingual OCR and Analytics for Noisy Unstructured Text Data was held in conjunction with ICDAR2011 in Beijing, China. The scope of `Multilingual OCR' is defined to include systems that are capable of reading more than one language in the same document, as well as one-language-per- document systems that can be easily retargeted to new languages. The proposed workshop will provide a forum for technical discussions on three important themes: i) recent progress in the field and promising new techniques , ii) attempts to identify and address 'hard' open research problems, and iii) performance evaluation of multilingual OCR systems.
With its emphasis on multi-lingual documents and encouragement of "work in progress" reports, the workshop is intended to be complementary to the main conference. In order to ensure that there is no conflict between submissions to the main conference and to the workshop, the paper submission deadline for the workshop will be after the decisions for the main conference have been made on July 1, 2017. The main motivation for this Workshop is to
Encourage "work in progress" manuscripts and preliminary ideas as well as mature work
Encourage descriptions of large collaborative projects
Encourage papers discussing at least 2 languages/scripts
Emphasize public access of the datasets when applicable
Encourage groups to demonstrate working systems by providing additional time after each talk
Organize group papers by scripts and languages to contrast methods and compare results
Ensure full participation of attendees by moderators encouraging plenty of Q/A
Hold discussion periods at the end of each session to summarize "take home messages" based on the papers presented and the Q/A
The workshop will also put out a call for multilingual OCR demo systems from researchers allowing them to submit applications for live demonstration of their work. A subset of the program committee will screen the applications for appropriateness to this workshop and approve demos for presentation.
The topics that will be addressed by this Workshop are:
Proven Methodologies for OCR: Efficacy of existing methodologies for Latin script to other scripts (HMMs, Neural networks etc.)
