Скачать презентацию Pavel Praks IEEE CBMI 2008 IEEE ICIP
a4142b70d8c630317505b92a2d1f54d9.ppt
- Количество слайдов: 14
Скачать презентацию Pavel Praks IEEE CBMI 2008 IEEE ICIP
Скачать презентацию Pavel Praks IEEE CBMI 2008 IEEE ICIP
Pavel Praks @ IEEE CBMI 2008 IEEE ICIP 2008 NIST Trec. Vid 2008 KEG, KIZI VŠE Praha, 18. 12. 2008
CBMI-2008 Sixth International Workshop on Content-Based Multimedia Indexing 18 -20 th June, 2008, London, UK ~ 90 papers http: //cbmi 08. qmul. net/techprog_detail. php Interesting paper: Web-scale System for Image Similarity Search: When the Dreams are Coming True David Novak, Michal Batko and Pavel Zezula Project MUFIN: http: //mufin. fi. muni. cz/
CBMI 08 Demo: Integration of an image retrieval Latent Semantic Indexing tool to a web service P. Praks (UEP) and K. Chandramouli (QMUL) The aim of the demonstration is to show our experience with an integration of a LSI-based image retrieval algorithm, which has been coded in Matlab, to a web-based multimedia retrieval framework. K-Space Content Retrieval System http: //kspace. qmul. net: 8080/cbmi-demo/
LINEAR ALGEBRA FOR VISION-BASED SURVEILLANCE IN HEAVY INDUSTRY CONVERGENCE BEHAVIOR CASE STUDY CBMI 2008, QMUL Pavel Praks 1, 2, Vojtěch Svátek 1, Jindřich Černohorský 2 1 University of Economics, Prague, Dept. of Information and Knowledge Engineering, Prague, Czech Republic K-Space No. E, www. k-space. eu 2 VŠB – Technical University of Ostrava, Czech Republic; Faculty of Electrical Engineering and Computer Science
Description of work The surveillance application aims at improving the quality of technology via modelling human expert behaviour in the coking plant Arcelor. Mittal Ostrava, the Czech Republic. Video data on several industrial processes are captured by means of a CCD camera and classified by using Latent Semantic Indexing (LSI) with the respect to etalons classified by an expert. We also study the convergence behavior of proposed partial eigenproblem-based dimension reduction technique and its ability for knowledge acquisition.
Case studies of images taken from the coking plant Arcelor. Mittal Ostrava, CZ Subjective evaluation of LSI-based results related to these two different settings: Experiment 1: k = 8 largest singular values is assumed Experiment 2: k = 45 largest singular values were used for LSI. For each case, query image represents a different industrial process. Image retrieval results are presented by decreasing order of similarity. The query image is situated in the upper left corner. (The similarity of the query image and the retrieved image is written in parentheses. ) In order to achieve well arranged results, only 9 most significant images are presented.
LSI Image retrieval results: Case D, k=8 The query image includes the detailed view of coke. All of the 8 most similar images are related to the same topic.
LSI Image retrieval results: Case D, k=45 The retrieved images are not related to the same topic at all.
Properties of LSI Image retrieval
2008 IEEE International Conference on Image Processing October 12– 15, 2008 • San Diego, California, U. S. A. ~ 800 papers (!!) http: //www. icip 08. org/ Interesting papers: Vacura M. , Svatek V. , Saathoff C. , Franz T. , Troncy R. : Describing Low-Level Image Features Using The COMM Ontology. Maleki A. , Shahram M. , Carlsson G. : A Near Optimal Coder For Image Geometry With Adaptive Partitioning. (Stanford University, http: //wwwvideo. eecs. berkeley. edu/Proceedings/ICIP 2008/pd fs/0001061. pdf (wedgelets))
P. Praks, E. Izquierdo, R. Kučera: The sparse image representation for automated image retrieval. IEEE ICIP 2008. We keep the memory limit of the decomposed data by a statistical model of the sparse data. The effectiveness of our novel approach is demonstrated by the large scale image similarity task of the NIST Trec. Vid 2007 benchmark. http: //www-video. eecs. berkeley. edu/Proceedings/ICIP 2008/pdfs/0000025. pdf
NIST TREC Video Retrieval Evaluation Online Proceedings 17 -18 November 2008, Gaithersburg, Maryland, USA. http: //www-nlpir. nist. gov/projects/tvpubs/tv. pubs. org. html UEP @ K-Space team: P. Wilkins, D. Byrne, Gareth J. F. Jones, H. Lee, G. Keenan, K. Mc. Guinness, N. E. O'Connor, N. O'Hare, A. F. Smeaton, T. Adamek, R. Troncy, A. Amin, R. Benmokhtar, E. Dumont, B. Huet, B. Merialdo, G. Tolias, E. Spyrou, Y. Avrithis, G. Th. Papadopoulous, V. Mezaris, I. Kompatsiaris, R. Mörzinger, P. Schallauer, W. Bailer, K. Chandramouli, E. Izquierdo, L. Goldmann, M. Haller, A. Samour, A. Cobet, T. Sikora, P. Praks, D. Hannah, M. Halvey, F. Hopfgartner, R. Villa, P. Punitha, A. Goyal, J. M. Jose, "K-Space at TRECVid 2008", TREC Video Retrieval Evaluation, November, 2008 http: //www-nlpir. nist. gov/projects/tvpubs/tv 8. papers/kspace. pdf
P. Praks @ Université Libre de Bruxelles: Modelování degradace Weibullovým rozdělením Výzkum ve spolupráci s prof. Pierre-Etienne LABEAU, Service de Métrologie Nucléaire, Université Libre de Bruxelles, Brusel, Belgie. Vstupem vyvíjeného software jsou doby do poruchy (+ "stopping times" pro cenzorovaná data). Implementovány 3 ekonomické modely popisující různé strategie údržby (ABAO, AGAN, BAGAN). Populární Weibull-modely nejsou vždy vhodné pro aproximaci vanové křivky:
Děkuji Vám za pozornost!
Скачать презентацию Pavel Praks IEEE CBMI 2008 IEEE ICIP
a4142b70d8c630317505b92a2d1f54d9.ppt