CUBSR&D PortfolioVenuGovindarajuDistinguished Professor, SUNY Buffalo
OverviewUnique CapabilitiesSponsorsTechnology Transfer RecordProjectsBiometricsDocument Recognition and RetrievalSecurityPeople
Unique CapabilitiesFaculty strengths in multiple disciplinesBehavioral Sciences, Social IssuesComputer Vision, VisualizationChemical and Biological SensorsPattern Recognition, Machine LearningSmart Environments, Pervasive ComputingSpectroscopySolid record of transferring of technology to field Large pool of  current PhD students (10)Growing pool of PhD alumni in industry (25)Several current projects with industry
Sponsors(last 5 years)ACIS, Buffalo, NYApplied Media Analysis, College Park, MDBBN Technologies, Cambridge, MABuffalo Computer Graphics, Blasdell, NYCUBRC, Cheektowaga, NYFujitsu, Sunnyvale, CAHP Labs, IndiaHealth Transaction Network, Williamsville, NYMatrix, Niagara Falls, NYUltra-Scan, Amherst, NYArmy Research Labs
Defense Intelligence Agency  (DIA)
Directorate of Central Intelligence (DCI)
National Endowment of Humanities (NEH)
National Science Foundation (NSF)
NYSTAR
Oishei FoundationTechnology Transfer4 US patents awardedBiometric convolution; Handwriting comparisons; Diagnosis of physiological states; Handwriting recognition4 US patents pendingFingerprint hashing; Deceit and verity; Document classification; Document Imagecapture Licensed technology to industryKyos Systems, Xact Data, Buffalo Computer Graphics, Lockheed MartinUSPS 1999 Annual Report"USPS issued a contract to SUNY Buffalo to develop the handwriting recognition technology. ….. an estimated 400 million pieces of mail were automatically routed ….. saved the Postal Service at least $90 million in its first year in the field.
BIOMETRICSCUBS
AnthropometricsSKIN TONE
Fingerprint IndexingEnrollment PhaseSearching Phase
CryptographyCancelable Biometrics
SensorsSkin spectroscopy and “liveness”
Fusion
Spoofing in Multimodal Systems
Facial Passwords
Q: The suspect is male1st IterationPruned SetQ: The suspect has a beard2nd IterationPruned SetQ: The suspect wears spectacles3rd IterationSUSPECTSoft BiometricsSemantic Face RetrievalOriginal Set
Unobtrusive People TrackingFreedom from  Continuous SurveillanceRECOGNIZEREASONEvolutionaryRecognitionRETRIEVEDid Bob and Frank meet at the library yesterday?Given building map, occupants, schedules, sensor locations
Soft BiometricsCrash scene analysis
DOCUMENT RECOGNITION AND RETRIEVALCUBS
Document Enhancement,
Multilingual Information RetrievalQ: Can we have a searchable archive of world’s newspapers?Q: All newspapers in any language translated to a common language?Central RepositorySearchable database(digitized)
Smart EMR
Handwriting Forensics
???Web SecurityVerify humannessMACHINES   FAILSynthetic handwriting generator poses questions in varying writing stylesEnglish!HUMANS  SUCCEED
securityCUBS
Soft BiometricsExpressions+=AU 6AU 12AU 6 & 12++=AU 1AU 2AU 4AU 1, 2 & 4++=AU 1AU 4AU 15AU 1, 4 & 15FACSFEARHAPPYANGERSADFEARHAPPYSADFacial Expression Manifold
Gaze Tracking
Soft BiometricsEmotional States
Deceit and Verity
PEOPLECUBS
PeopleFacultyFrank BrightSUNY Distinguished ProfessorBiological, Chemical SensorsVenuGovindarajuSUNY Distinguished ProfessorMachine LearningDirectorMark FrankProfessorBehavioral SciencesRaymond FuAssistant ProfessorComputer Vision, VisualizationAlex CartwrightProfessorSpectroscopy, PhotonicsBharat JayaramanProfessorCyber Physical Systems

CUBS expertise

Editor's Notes

  • #19 Some more detail concerning the impact of ruled line removal on word recognition:We extracted all the test word images from lined pages and measured the top choice recognition performance. Here are the numbers: -- Total word images in test set : 848 from a total of 274 pages. Of these: -- Number of word images from pages with ruled lines: 460, from 146 lined pages. -- The ratio of words and pages with ruled lines in the 34 PAW data set: 460/848 = 54.25% (word), 146/274=53.28% (pages).Recognition performance on words from lined pages: -- Top1: Earlier: 318/460 = 69.13% Now: 349/460 = 75.87% The ruled line removal improves the word recognition for top 1 by 6.74% (evaluated on words from lined pages). Overall improvement for top 1 is by 4.13% (evaluated using test set including all word images from lined or non-lined pages - which we had reported earlier).Also the PAW recognizer is a straightforward implementation using a k-nearest neighbor classifier. The features used are CUBS Gradient, Structure and ConcavityFeatures. The classifier is a very simple implementation that can be improved and its purpose was for testing the effectiveness of our features.