ECE160 / CMPS182 MultimediaNational Research PrioritiesMultimedia RecognitionWide-area SurveillanceSlide 5Slide 6How to Organize these Photos?Image Organization & RetrievalEXTENTTM = contEXT + contENTEXTENTTMAugmented ImagesContext from Space/TimeExample of Using Three Pieces of InformationMaui Sunsets can be obtained from Space/TimeUse content for verificationUse content to transfer metadataSummarize of the exampleAre They Similar?Slide 19Are They Similar?Conveying PerceptionSlide 22Keyword RetrievalSlide 24Slide 25Slide 26Slide 27Recogintion of ContentRecognitionSlide 30Slide 31Web 1.0 vs. Web 2.0Web 2.0FotofitiECE160 Spring 2009 Lecture 20 1ECE160 / CMPS182MultimediaLecture 20: Spring 2009Image Recognition and RetrievalECE160Spring 2009 Image Recognition 2National Research PrioritiesEnergy TechnologiesFuel efficient enginesReplacement energy to fossil fuelsLighter, longer-duration batteriesBioengineering/BioinformaticsGenes diseaseDisease medicineSearch with Multimedia Content Video surveillance Photo interpretationECE160Spring 2009 Image Recognition 3Multimedia RecognitionVideo surveillancePhoto interpretationECE160Spring 2009 Image Recognition 4Wide-area Surveillanceadvertisement of objectvideo.comECE160Spring 2009 Image Recognition 5Surveillance Scenarios(1) Intrusion Detection(1) Intrusion Detection (2) Passenger Screening(2) Passenger Screening(3) Perimeter Monitoring(3) Perimeter MonitoringZZUse biometric facial recognition to identify individuals of interest through existing closed circuit TV surveillanceMonitor and alert on tailgating, loitering, exit/closed entry, other unauthorized accessObject tracking and biometric facial recognition to determine vehicles and humans exhibiting suspicious behavior(4) Unattended Baggage(4) Unattended BaggageCopyright © 2004 Proximex Corp.Copyright © 2004 Proximex Corp.Identify unattended baggage (or other objects) left for long periods of timeECE160Spring 2009 Image Recognition 6Multimedia RecognitionVideo surveillancePhoto interpretationECE160Spring 2009 Image Recognition 7How to Organize these Photos?ECE160Spring 2009 Image Recognition 8Keyword-based Manual labeling is subjective, cumbersomeThe aliasing problemContent-basedPromising for general semantics: outdoor, landscape, flowers, people, etc. Not enough for w h-queries (where, who, when, or what)Image Organization & RetrievalECE160Spring 2009 Image Recognition 9EXTENTTM = contEXT + contENTContext Spatial (location)Temporal Social OthersContent Perceptual features, such as color, texture, and shapeHolistic features and local featuresECE160Spring 2009 Image Recognition 10EXTENTTMContentSpatial TemporalSocial OthersECE160Spring 2009 Image Recognition 11Augmented Images+Cameraphones with high-quality lens can record location, time, camera parameters, and voice=ECE160Spring 2009 Image Recognition 12Context from Space/TimeGPS or CellID data Into place names Time-based groupingInto meaningful “events”From place names and timeTime of dayWeatherECE160Spring 2009 Image Recognition 13Example of Using Three Pieces of Information ContentSpatial TemporalECE160Spring 2009 Image Recognition 14Maui Sunsetscan be obtained from Space/TimeECE160Spring 2009 Image Recognition 15Use content for verificationECE160Spring 2009 Image Recognition 16Use content to transfer metadataECE160Spring 2009 Image Recognition 17Summarize of the exampleDerived from ContextDerive time of the dayObtain weatherVerify contentUse of ContentVerify contextTransfer contextMuch more…ECE160Spring 2009 Image Recognition 18Are They Similar?ECE160Spring 2009 Image Recognition 19Are They Similar?ECE160Spring 2009 Image Recognition 20Are They Similar? In terms of what?What is the user’s perception?ECE160Spring 2009 Image Recognition 21Conveying Perception Image DatabasesConveyed via ExamplesUse a sunset picture (or pictures) to find more sunset imagesWhere does the perfect example come from?ECE160Spring 2009 Image Recognition 22Conveying Perception Internet SearchesConveyed via KeywordsECE160Spring 2009 Image Recognition 23Keyword Retrieval ProsA user-friendly paradigmConsAnnotation is a laborious processAnnotation quality can be subparAnnotation can be subjectiveSynonymsECE160Spring 2009 Image Recognition 24Conveying Perception Image DatabasesConveyed via ExamplesUse a sunset picture (or pictures) to find more sunset imagesWhere does the perfect example come from?ECE160Spring 2009 Image Recognition 25Are They Similar?ECE160Spring 2009 Image Recognition 26Are They Similar?ECE160Spring 2009 Image Recognition 27Are They Similar? In terms of what?What is the user’s perception?ECE160Spring 2009 Image Recognition 28Recogintion of ContentECE160Spring 2009 Image Recognition 29RecognitionECE160Spring 2009 Image Recognition 30RecognitionECE160Spring 2009 Image Recognition 31clouds vs. wavesECE160Spring 2009 Image Recognition 32Web 1.0 vs. Web 2.0UserUser UserContentContent (Image/Video)ContentContentContentContentContentContent (text)ContentContentContentContentContentUserUserUser UserECE160Spring 2009 Image Recognition 33Web 2.0 Content + Users + InteractionsCollect rich, organized contentAttract users & interactions To provide metadataTo provide new contentImprove search qualityWith new metadata and dataVia social-network structureECE160Spring 2009 Image Recognition 34User managementUser managementPhoto uploadingSingle/multipleUpload wizardPhoto uploadingBuilding social networkSocial networksEvent managementEvent managementPhoto searchPhoto searchMetadata collectionMetadata fusionImage annotationMetadata collection - contextual - content Metadata fusion Annotate photos External functionalitiesInternal
View Full Document