Search

Search Funnelback University

Search powered by Funnelback
1 - 35 of 35 search results for news |u:mi.eng.cam.ac.uk
  1. Fully-matching results

  2. EARS Project

    mi.eng.cam.ac.uk/research/projects/EARS/ears_summary.html
    7 Jul 2003: The aim is to improve and develop new, generally applicable, techniques for speech recognition. ... Apart from new functionality, recipes'' and infrastructure will be developed for a number of standard tasks.
  3. RECENT ADVANCES IN BROADCAST NEWS TRANSCRIPTION D.Y. Kim, G. ...

    mi.eng.cam.ac.uk/reports/svr-ftp/kim_asru2003.pdf
    25 Sep 2003: New trigram LM was used that is presented in section7. the necessary pronunciation statistics were obtained from align-ment of the Switchboard and Broadcast News training corpora.The SPRON dictionaries were ... Broadcast News Data. Acoustic training data.
  4. Bin Jia, Khe Chai Sim et al: CU-HTK RT03 ...

    mi.eng.cam.ac.uk/research/projects/EARS/pubs/jia_rt03s.pdf
    23 Jun 2003: Language Model. • Sources of data (using LDC character-to-word segmentor)– Acoustic training data (modifier Kneser-Ney)– News corpora: TDT[2,3,4], China Radio, People’s Daily, Xinhua (Good-.
  5. RECONSTRUCTION OF SCULPTURE FROM UNCALIBRATED IMAGE PROFILES Roberto…

    mi.eng.cam.ac.uk/reports/svr-ftp/kykwong_icip2001.pdf
    5 Dec 2003: The circular motion will generate a web of contour gen-erators around the object, which can be exploited for reg-istering any new arbitrary general view. ... 2). j0. e j1 e j2. e. Fig. 2. Three views from circular motion provide 6 outerepipolar tangents
  6. Kim et al.: 2003 CU-HTK BN-E Systems 2003 CU-HTK ...

    mi.eng.cam.ac.uk/research/projects/EARS/pubs/kim_rt03s.pdf
    24 Jul 2003: Kim et al.: 2003 CU-HTK BN-E Systems. 2003 CU-HTK Broadcast News English SystemDevelopment. ... B CNN shows transcription 1999-2001 66C TDT4 closed captions 2D broadcast news acoustic training transcriptions 1997-1998 2.
  7. Model-Based Hand Tracking Using anUnscented Kalman Filter B. Stenger, …

    mi.eng.cam.ac.uk/reports/svr-ftp/stenger_bmvc01.pdf
    12 May 2003: Sequential Monte CarloMethods in Practice. Statistics for Engineering and Information Science Series.Springer-Verlag, New York, 2001. ... J. Julier, J. K. Uhlmann, and H. F. Durrant-Whyte. A new approach for filteringnonlinear systems.
  8. Diarisation for RT-03s at Cambridge University Sue Tranter, Kai ...

    mi.eng.cam.ac.uk/research/projects/EARS/pubs/tranter_rt03sdiary.pdf
    22 Jul 2003: Some inconsistency in the way pre-recorded announcements (e.g. ’Thisis the news from ABC’) are transcribed in the reference. ... BNEWS - Segmentation - System Description. Segmentation is based on the CUHTK Hub-4 1998 10xRT STT system.Modifications
  9. A Practical Method for Estimation of PointLight-Sour ces Martin ...

    mi.eng.cam.ac.uk/reports/svr-ftp/weber_bmvc2001.pdf
    29 Nov 2003: Isard. ActiveContours. Springer, New York, 1998. [2] R. Cipolla andP. Giblin. ... 12] R.J.Woodham.Gradientandcurvaturefrom thephotometric-stereomethod,includ-ing local confidenceestimation.JOSA, 11:3050–3068,1994. [13] P.L. WorthingtonandE.R.
  10. Ongoing Experiments with Fisher Data Ricky Chan, Gunnar Evermann ...

    mi.eng.cam.ac.uk/research/projects/EARS/pubs/chan_stthomas03.pdf
    10 Dec 2003: New LMs: Eval03 with CU-HTK P1-P2 System. Overall Swbd Fisher Male Femaleh5train03b LM03 24.6 28.7 20.2 25.7 23.5h5train03b LM03Fi 23.9 28.2 ... Chan et al.: Ongoing Experiments with Fisher Data. New LMs: Eval02 with CU-HTK P1-P2 System.
  11. DESIGN OF FAST LVCSR SYSTEMS G. Evermann & P.C. ...

    mi.eng.cam.ac.uk/reports/svr-ftp/evermann_asru2003.pdf
    23 Sep 2003: For the 2003 evaluation datawas taken from the new LDC Fisher collection1 and Switch-board II phase 5. ... More details on the effectiveness of these tech-niques on Broadcast News can be found in [8].
  12. Evermann, Kim, Wang, Woodland et al.: CU-HTK Fast System ...

    mi.eng.cam.ac.uk/research/projects/EARS/pubs/evermann_rt03s.pdf
    23 Jun 2003: 2003 System structure 10xRT CTS. • Automatic segmentation. • Use new models from full system. •
  13. Model-Based 3D Tracking of an Articulated Hand B. Stenger ...

    mi.eng.cam.ac.uk/reports/svr-ftp/stenger_cvpr01.pdf
    12 May 2003: Springer-Verlag,New York, 2001. [12] D. M. Gavrila and L. S. Davis. ... 17] A. H. Jazwinski. Stochastic Processes and Filtering Theory.Academic Press, New York, 1970.
  14. Optimisation of Fast LVCSR Systems Gunnar Evermann, Phil Woodland ...

    mi.eng.cam.ac.uk/research/projects/EARS/pubs/evermann_stthomas03.pdf
    10 Dec 2003: New 10xRT system. Changes:. • Faster P1 configuration. • Use SPron model for lattice generation (about 10% faster). •
  15. Automatic 3D Modelling of Architecture Anthony Dick� Phil Torr� ...

    mi.eng.cam.ac.uk/reports/svr-ftp/dick_bmvc00.pdf
    5 Dec 2003: A projection matrix is then estimated linearly from the 2D-3D cor-respondences, and new structure is triangulated from 2D-2D matches without previousstructure. ... is computed by subtracting the contributions from the invalid. boundary segments (shown as
  16. Metadata Extraction at Cambridge University Sue Tranter, Marcus…

    mi.eng.cam.ac.uk/research/projects/EARS/pubs/tranter_earsjan03.pdf
    8 Jul 2003: C.Woodland & S.J.YoungSegment Generation & Clustering in the HTK Broadcast News TranscriptionSystemProc. ... 1998 DARPA Broadcast News Transcription and Understanding Workshop, pp.133-137 (Lansdowne, VA, Feb.
  17. Learning a Kinematic Priorfor Tr ee-Based Filterin g A. ...

    mi.eng.cam.ac.uk/reports/svr-ftp/thayananthan_bmvc03.pdf
    23 Nov 2003: Currently, mosttrackers useparticlefiltering [7], in which it is essentialto be abletosamplefrom the prior to generate new hypotheses.With the increasein computationalpower onemay also considerhandlingambiguoussituationsby treatingtracking as
  18. Woodland et al.: English CTS Systems 2003 CU-HTK English ...

    mi.eng.cam.ac.uk/research/projects/EARS/pubs/woodland_rt03s.pdf
    24 Jul 2003: After eval03 found SPron lattices from scratch (new word lattices/modelmarked lattice MPE regen) helps by only another 0.1% absolute. ... variable number of Gaussians per state: 0.3% abs. • new MPE lattice generation/regeneration procedure: 1.2% abs.
  19. Filterin g Using a Tr ee-BasedEstimator B. Stenger� A. ...

    mi.eng.cam.ac.uk/reports/svr-ftp/stenger_iccv03.pdf
    23 Nov 2003: We have testedthe new tracking methodon sequencesinvolving clutter in the background together with non-rigidhandmotion. ... Barrow, J. M. Tenenbaum, R. C. Bolles, and H. C.Wolf. Parametriccorrespondence and chamfer matching:Two new techniques for
  20. Interactions Between Diarisation and STT Sue Tranter1, Kai Yu1, ...

    mi.eng.cam.ac.uk/research/projects/EARS/pubs/tranter_rt03sinter.pdf
    22 Jul 2003: 1] 10xRT BN RT-03 system with new LM excluding days from bndidev03 data set. • ... BN - Mixing-and-matching - the details. • Removing Commercials. – ( NONE / CUED EVAL / CUED TDT4 / PERFECT )– use the output as the new UEM input to the
  21. Reconstruction of Surfaces of Revolutionfrom Single Uncalibrated…

    mi.eng.cam.ac.uk/reports/svr-ftp/kykwong_bmvc02.pdf
    5 Dec 2003: InProc. Conf. Computer Vision andPattern Recognition, pages 123–129, New York, NY, June 1993.
  22. Bayesian Stochastic Mesh Optimisation for 3DReconstruction George…

    mi.eng.cam.ac.uk/reports/svr-ftp/vogiatzis_bmvc03.pdf
    23 Nov 2003: Vertex v is replaced by verticesv1 andv2 and new edgee joining them isintroduced. ... The reason it fails is that erroneous initial verticescannot be removed and new areas of the surface cannot be added.
  23. Recent Experiments with HTK Broadcast Newsand Conversational…

    mi.eng.cam.ac.uk/research/projects/EARS/pubs/woodland_earsjan03.pdf
    17 Jan 2003: Some new techniques including:– MPE/MMI-MAP– discriminative SAT– variable number of Gaussians/state. • ... First release from this tree was HTK3.2, December ’02, new features include:. –
  24. iccv.dvi

    mi.eng.cam.ac.uk/reports/svr-ftp/drummond_iccv2001.pdf
    20 Oct 2003: This gives new values for µ2 and C2 which take intoaccount the optimal pose of component 1. ... The process can be re-peated, propagating the new statistics for component 2 intothe next component up the chain until the statistics for allcomponents have
  25. Shape Context and Chamfer Matching in Cluttered Scenes A. ...

    mi.eng.cam.ac.uk/reports/svr-ftp/thayananthan_cvpr03.pdf
    12 May 2003: C. Bolles, and H. C.Wolf. Parametriccorrespondence and chamfer matching:Two new techniquesfor imagematching. ... Rangarajan. A new point matchingalgo-rithm for non-rigid registration.ComputerVision and ImageUnderstanding, 2003.
  26. Towards a Complete Dense Geometric andPhotometric Reconstruction…

    mi.eng.cam.ac.uk/reports/svr-ftp/weber_bmvc2002.pdf
    29 Nov 2003: JOSA, 11:3050–3068, 1994. [16] P.L. Worthington and E.R. Hancock. New constraints on data-closeness and needlemap consistency for shape-from-shading.
  27. Initialisation and Termination of Active Contour Level-Set Evolutions …

    mi.eng.cam.ac.uk/reports/svr-ftp/weber_vlsm2003.pdf
    29 Nov 2003: The firstterm (. H) - ) K ) accounts for the fact that as changes, theevaluation of - has to be translated to the new position of theinterface (i.e.
  28. ASRU03.dvi

    mi.eng.cam.ac.uk/reports/svr-ftp/wang_asru2003.pdf
    25 Sep 2003: In Fig. 1, “INPUT” indicates that these transformsare applied to the feature/model space when calculating the occu-pancies for new linear transforms, while “PARENT” means thosetransforms are also applied when
  29. paper.dvi

    mi.eng.cam.ac.uk/reports/svr-ftp/liu_asru2003.pdf
    25 Sep 2003: Given the currentvalue of the criterion, the new value is estimated as the old value.
  30. IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. XX, NO. Y, ...

    mi.eng.cam.ac.uk/reports/svr-ftp/kykwong_tip04.pdf
    18 Sep 2003: XX, NO. Y, MONTH YEAR. overcome by incorporating new views from arbitrary general motion. ... rotation axis, the horizon and the special vanishing point, respectively, under this new camera.
  31. A Sparse Probabilistic Learning Algorithm for Real-Time Tracking…

    mi.eng.cam.ac.uk/reports/svr-ftp/williams_iccv03.pdf
    23 Nov 2003: New methods in Wiener filtering. In Proc. of theFirst Symposium on Engineering Applications of RandomFunction Theory and Probability.
  32. ivc2105.dvi

    mi.eng.cam.ac.uk/reports/svr-ftp/kykwong_ivc02.pdf
    5 Dec 2003: M. Riseman (Eds.), Computer Vision Systems, AcademicPress, New York, 1978, pp.
  33. CAMBRIDGE UNIVERSITYENGINEERING DEPARTMENT Automatic Transcription…

    mi.eng.cam.ac.uk/reports/svr-ftp/hain_tr465.pdf
    18 Dec 2003: This canonical model should be more compact and amenable to being transformed toa new speaker, or acoustic condition, than standard speaker independent (SI) systems. ... 50000 most frequent words occurring in 204 million words (MW)of Broadcast News (BN)
  34. 19 May 2003: Thus, only z offsets involve calculating a new reslice of sweep B: x and yoffsets can be evaluated by sliding the same sweep B reslice over the sweep A reslice.
  35. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL.…

    mi.eng.cam.ac.uk/reports/svr-ftp/kykwong_pami03.pdf
    5 Dec 2003: IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, VOL. XX, NO. Y, MONTH YEAR 1. Camera Calibration. from. Surfaces of Revolution. Kwan-Yee K. Wong, Paulo R. S. Mendonca and Roberto Cipolla. Kwan-Yee K. Wong (corresponding author) is
  36. An Investigation into the Interactions betweenSpeaker Diarisation…

    mi.eng.cam.ac.uk/reports/svr-ftp/tranter_tr464.pdf
    9 Oct 2003: telephone speech (CTS) and both CUED and MIT-LL for broadcast news (BN). ... 2 Diarisation 10. Data Set Scheme Broadcaster Audio Adverts ’News’Removed Removed Removed.

Refine your results

Related searches for news |u:mi.eng.cam.ac.uk

Search history

Recently clicked results

Recently clicked results

Your click history is empty.

Recent searches

Recent searches

Your search history is empty.