17. Neural Ctrl-F : Segmentation-free query-by-string word spotting in handwritten manuscript collec-tions

Authors:Tomas Wilkinson(*), Jonas Lindstr¨om(1), Anders Brun(*) (*) CBA

(1) Dept. of History, UU

In Proceedings:IEEE International Conference on Computer Vision (ICCV), 2017, pp. 4433-4442 Abstract: In this paper, we approach the problem of segmentation-free query-by-string word spotting for handwritten documents. In other words, we use methods inspired from computer vision and machine learn-ing to search for words in large collections of digitized manuscripts. In particular, we are interested in historical handwritten texts, which are often far more challenging than modern printed documents. This task is important, as it provides people with a way to quickly find what they are looking for in large collec-tions that are tedious and difficult to read manually. To this end, we introduce an end-to-end trainable model based on deep neural networks that we call Ctrl-F-Net. Given a full manuscript page, the model simul-taneously generates region proposals, and embeds these into a distributed word embedding space, where searches are performed. We evaluate the model on common benchmarks for handwritten word spotting, outperforming the previous state-of-the-art segmentation-free approaches by a large margin, and in some cases even segmentation-based approaches. One interesting real-life application of our approach is to help historians to find and count specific words in court records that are related to women’s sustenance activities and division of labor. We provide promising preliminary experiments that validate our method on this task.

18. Distance between vector-valued fuzzy sets based on intersection decomposition with applications in object detection

Authors:Johan ¨Ofverstedt(*), Nataˇsa Sladoje(*,1), Joakim Lindblad(*,1) (*) CBA

(1) Mathematical Institute of Serbian Academy of Sciences and Arts, Belgrade, Serbia

In Proceedings: Mathematical Morphology and its Applications to Signal and Image Processing. ISMM 2017. Lecture Notes in Computer Science, vol. 10225, pp. 395–407

Abstract:We present a novel approach to measuring distance between multi-channel images, suitably rep-resented by vector-valued fuzzy sets. We first apply the intersection decomposition transformation, based on fuzzy set operations, to vector-valued fuzzy representations to enable preservation of joint multi-channel properties represented in each pixel of the original image. Distance between two vector-valued fuzzy sets is then expressed as a (weighted) sum of distances between scalar-valued fuzzy components of the trans-formation. Applications to object detection and classification on multi-channel images and heterogeneous object representations are discussed and evaluated subject to several important performance metrics. It is confirmed that the proposed approach outperforms several alternative single- and multi-channel distance measures between information-rich image/object representations.

2. Discrete convolution operators, the Fourier transformation, and its tropical counterpart: the Fenchel transformation

Author:Christer O. Kiselman(*) (*) CBA

In Proceedings:3rd EAUMP Conference: Advances in Mathematics and its Applications, Kampala, Uganda:

Makerere University, p. 7-28

3. Lars H¨ormander - some early memories (Chinese) Author:Christer O. Kiselman(*)

(*) CBA

In Proceedings:Mathematical Advances in Translation, Vol. 36, no 2, p. 120-121 Comment:Translation into Chinese of the English original translated in 2015

4. Watching solar eclipses 1945 - 2017 Author:Christer O. Kiselman(*) (*) CBA

Journal:Bulletin of the Swedish Mathematical Society, no October, p. 17-21 Comment:Science popularization

5. Br˚ak och spr˚ak – vad som ¨ar f¨ornuftigt och logiskt Authors:Christer O. Kiselman(*), Hania Uscka- Wehlou(1) (*) CBA

(1) Dept. of Mathematics, UU

Journal:N¨amnaren : tidskrift f¨or matematikundervisning, no 1, p. 45-49 Comment:Science popularization

6. Falska v¨anner, vassa vr˚ar och spr˚akliga f¨allor

Authors:Christer O. Kiselman(*), Hania Uscka-Wehlou(1) (*) CBA

Dept. of Mathematics, UU

Journal:N¨amnaren : tidskrift f¨or matematikundervisning, no 2, p. 43-51 Comment:Science popularization

7. CBA Annual Report 2016

Editors: Marine Astruc, Gunilla Borgefors, Filip Malmberg, Lena Nordstr¨om, Ingela Nystr¨om, Leslie Solorzano, Robin Strand

Publisher:Centre for Image Analysis, 110 pages

7 Activities

This year, we were part of organising the first Swedish Symposium on Deep Learning, a Sum-mer school for PhD students in Novi Sad, Serbia, and a Workshop together with a Korean Company. The Symposium was very well attended, as this is a very “hot” subject at present.

We are often invited to give seminars outside CBA, this year in Uppsala, Denmark, Russia, Serbia, Canada, and USA. Something we are really proud of is our own longstanding internal seminar series, with one or two 30–45 minute seminars every Monday afternoon. This year, we had 39 seminars, nine of which were given by guests. The average number of attendants was 23, significantly higher than the previous years.

As usual, we attended many national and international meetings, where we presented our work as invited speaker or giving oral or poster presentations of reviewed papers. We also gave presentations at non-reviewed meetings. Attending national and international meetings is inspiring and necessary to be part of the scientific community.

We had an unusual number of visiting scientists, staying for longer periods. Professor Heung-Kook Choi from Inje University Korea stayed for a full sabbatical year. This was a revisit, as he was a PhD student at CBA 1990–96 supervised by Ewert Bengtsson. Another distinguished guest was Professor Douglas Hofstadter from Indiana University, USA who stayed for three months. Other visitors came from Finland, Estonia, Serbia, and Canada.

Other ways of being part of the international scientific community is working for profes-sional organisations, being Editors of scientific journals, serving in programme committees for international and national conferences, reviewing for international journals (which often goes undocumented), being members of dissertation committees, and functioning as evaluators of projects and positions.

(a) (b)

Figure 66: Wordcloud of the titles and abstracts of the internal (a) and external (b) seminars.

Seminarier – 9 Nov saknas data – tog median (för externa)   

   

   

   

0 5 10 15 20 25 30 35

16 Jan 23 Jan 30 jan 6 Feb 13 Feb 20 Feb 27 Feb 6 Mar:1 6 Mar:2 20 Mar:1 20 Mar:2 27 Mar 3 Apr 10 Apr 24 Apr 8 May 15 May:1 15 May:2 22 May 29 May:1 29 May:2 19 Jun 28 Aug 4 Sep 11 Sep 18 Sep 25 Sep 9 Oct 16 Oct 23 Oct 30 Oct 6 Nov 9 Nov 13 Nov 20 Nov 27 Nov 4 Dec 11 Dec 18 Dec

ExF-ExG ExF-InG InF-ExG InF-InG

Figure 67: Our own seminar series. Blue represents seminars given by CBA people, while red represents guest lecturers. The saturated part on the bars represents guest attendants. For one seminar data is missing, this is shown as a blank bar and is represented by the median value (in its category).

7.1 Conference organization

1. The First Swedish Symposium on Deep Learning Organisers:SSBA

Address:Royal Institute of Technology, Stockholm Brinellv¨agen 64, Stockholm.

Date:20170620–20170621

Comment: Anders Brun was in the organizing committee and also one of the founders of this symposium series.

2. Summer School on Image Processing (SSIP) 2017

Organisers:Faculty of Technical Sciences, University of Novi Sad, Serbia Address:Faculty of Technical Sciences, University of Novi Sad, Serbia Date:20170713–20170722

Comment:Joakim Lindblad was head of the Program Committee 3. CBA and JLK Inspection (Korea Company) Joint Workshop

Organisers:Ewert Bengtsson, Carolina W¨ahlby, Myeong-Jae Lee Address:CBA

Date:20171107–20171107

I dokument Annual Report 2017 Centre for Image Analysis Centrum f¨or bildanalys (sidor 86-89)