DAR 2018 Progam (PDF)

Shortlisted Papers invited for presentation at DAR 2018:
2, 3, 4, 8, 11, 12, 14, 16, 17, 18, 20, 21, 22, 23

Registration for the workshop is included as part of ICVGIP registration.
Please click here to register

Overview & Objective

The ability to understand the content and structure of documents as well as text that appears in the form of handwriting or scene text is critical for intelligent systems to operate and interact with humans. With advances from research in the past few decades, the area has moved from solutions that work with highly restricted data to real-world degraded noisy document images and those acquired using hand-held cameras or handwriting data captured by mobile devices. The fourth workshop on document analysis and recognition aims to provide a platform for researchers to present techniques for solving such problems and interact with their peers in academia and industry. The topics of interest include, but are not limited to:

  • Text Recognition, including from video sources
  • Document Layout Analysis and Understanding
  • Historical and Degraded Document Analysis
  • Multilingual Document Recognition
  • Document Image Compression
  • Camera-captured Document Analysis
  • Handwriting Recognition Technologies
  • Writer/style Identification, Adaptation
  • Document Image Retrieval
  • Datasets, Performance evaluation and Metrics

The workshop will be held on Tuesday, 18th December 2018 in conjunction with the tenth Indian Conference on Computer Vision, Graphics, and Image Processing (ICVGIP) at the International Institute of Information Technology, Hyderabad, India. We invite researchers to submit their unpublished papers describing high-quality and original research in various application areas of document image analysis and recognition. In addition to contributory research papers, the workshop will also host invited talks.

The manuscript for submission should be prepared according to Springer LNCS format. The length of the manuscript should not exceed 12 pages.

To download the LNCS format and guidelines, click here.

To submit a paper use the CMT link.

Program Highlights

DAR 2018 Progam (PDF)

Keynote Talk


A G Ramakrishnan (IISc Bangalore)


Title: Attention Feedback Vs DNN aka Lipi Gnani OCR Vs Google's Tessaract

Abstract: Google’s Tesseract OCR uses LSTM for recognition and gives a good output for most Indian languages. However, we show that a carefully designed and executed OCR with conventional SVM can easily match and even surpass it. Our Lipi Gnani OCR employs the patented Attention-Feedback technology for improving the binarization quality at the word level and segmentation effectiveness at the character level. We have created and made publicly available benchmark datasets of 241 Kannada printed pages from books printed over many decades, containing Halegannada, merged and split characters and also interspersed English words, together with ground truth in Unicode. We demonstrate superior performance of our Lipi Gnani Kannada OCR on the above 241 pages than that provided by Tesseract, even though we are yet to employ any dictionary-based correction. Similar trend is observed on Tulu, Konkani, Tamil and Sanskrit text printed in Kannada script. We have also created benchmark datasets of 1685 degraded word images and 4033 sub-word images obtained from old Kannada books, containing split and merged characters, respectively, both with ground truths in Unicode. We show that our preprocessing modules that improve the quality of the scanned gray images result in improved performance of the Tessearct OCR on the above two databases. Thus, we establish that there is ample scope for the fruitful marriage of sound image-processing techniques with the deep neural networks to further enhance the overall performance of DAR systems.

Paper Presentations

  • Both oral and poster sessions

For any queries please mail to: icdar2018dar@gmail.com

IMPORTANT DATES

  • Full manuscript due: 5th Nov 2018 12th Nov 2018
  • Notification for presentation
    (authors of shortlisted papers will be given a presentation slot at the DAR Workshop):

    27th November 2018
  • Post workshop author notification
    (Decision notification regarding acceptance of manuscript for publication in post workshop proceedings; detailed reviews visible to authors):
    30th December 2018
  • Camera ready paper: 21st Jan 2019


Advisory Committee

  • Bidyut Baran Chaudhuri (ISI Kolkata)
  • Santanu Chaudhury (CSIR-CEERI & IIT Delhi)
  • Umapada Pal (ISI Kolkata)
  • A G Ramakrishnan (IISc Bangalore)

Program Co-chairs

  • Suresh Sundaram (IIT Guwahati)
  • Gaurav Harit (IIT Jodhpur)

Program Committee

  • Ujjwal Bhattacharya (ISI Kolkata)
  • Utpal Garain (ISI Kolkata)
  • Prithvijit Guha (IIT Guwahati)
  • D S Guru (University of Mysore)
  • Thotreingam Kasar (Kaaya Tech)
  • G S Lehal (Punjabi University)
  • Anand Mishra (IISc)
  • Atul Negi (University of Hyderabad)
  • Swapan Parui (ISI Kolkata)
  • Shivakumara Palaiahnakote (University of Malaya)
  • Sitaram Ramachandrula (DXC Technology)
  • Partha Pratim Roy (IIT Roorkee)
  • R K Sharma (Thapar Institute of Engineering & Technology)

Workshop Details:

DAR 2018 will be a 100% participation, one day, single track workshop, featuring invited talks, oral and poster presentations.

Submission:

DAR 2018 invites previously unpublished and original work. The submissions will undergo blind review, where the referees remain anonymous for the authors throughout the process. Papers should not exceed 12 pages in the LNCS format.

Submissions can be made at the DAR 2018 CMT link.

Publication

Selected papers will be published by CCIS, Springer.

Event Schedule

12th Nov 2018

Full manuscript due

27th November 2018

Notification to authors of shortlisted papers for presentation at the workshop

30th December 2018

Decision Notification regarding publication of manuscript in post workshop proceedings;

Reviews available to the authors

21st January 2019

Camera ready manuscript due