Describing Images Farhadi etal CVPR 2009 No examples from these object categories were seen during training Describing Objects by their Attributes Farhadi etal CVPR 2009 Absence of typical attributes ID: 1026555
Download Presentation The PPT/PDF document "Describing Images Using Attributes" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
1. Describing Images Using Attributes
2. Describing ImagesFarhadi et.al. CVPR 2009
3. No examples from these object categories were seen during trainingDescribing Objects by their AttributesFarhadi et.al. CVPR 2009
4. Absence of typical attributes752 reports68% are correctFarhadi et.al. CVPR 2009
5. Presence of atypical attributes951 reports47% are correctFarhadi et.al. CVPR 2009
6. NormalitySaleh et. al. Object Centric Anomalty Detection by Attribute-Based Reasoning, CVPR13
7. Abnormal Object DatasetSaleh et. al. Object Centric Anomalty Detection by Attribute-Based Reasoning, CVPR13
8. Abnormality Prediction and RankingMethodAUCOne class SVM0.5980Two class SVM0.8657Graphical Model0.8703Our Model with surprise score0.9105Less AbnormalHigh AbnormalBased on Abnormality Score, we can classify an object as Normal vs. Abnormal.Also, using this score we are able to rank images based on how strange they look like.Saleh et. al. Object Centric Anomalty Detection by Attribute-Based Reasoning, CVPR13
9. Reasoning about Abnormality via AttributesSaleh et. al. Object Centric Anomalty Detection by Attribute-Based Reasoning, CVPR13
10. Describing ObjectsDetector inputStrongest category response with good overlapStrongest part response within each spatial binFarhadi et. al, Attribute-Centric Recognition for Cross-Category Generalization, CVPR10
11. Describing ObjectsLearn spatial correlations and co-occurrence Detector ResponsesTrue Value for Categories and Spatial PartsHas PartHas FunctionPose/ViewpointLatent “Root”Learned by EM in trainingFarhadi et. al, Attribute-Centric Recognition for Cross-Category Generalization, CVPR10
12. animalfunction: can bitefunction: can flypart: eyepart: footpart: headpart: legpart: mouthpart: tailpart: wingPose: objects_frontAnimalblc: eaglefunction: can bitefunction: can flyfunction: is predatorfunction: is carnivorouspart: eyepart: footpart: headpart: legpart: mouthpart: wingPose: extended_wingsPose: objects_front Describing Familiar ObjectsFarhadi et. al, Attribute-Centric Recognition for Cross-Category Generalization, CVPR10
13. Using Localized AttributesVehicleWheelAnimalLegHeadFour-leggedMammalCan runCan JumpIs HerbivorousFacing rightMoves on roadFacing rightFarhadi et. al, Attribute-Centric Recognition for Cross-Category Generalization, CVPR10
14. Relative (ours):More natural than insidecity Less natural than highway More open than street Less open than coast Has more perspective than highway Has less perspective than insidecityBinary (existing):Not naturalNot openHas perspectiveUsing Relative Attributes14Parikh, Grauman, Relative Attributes, ICCV 2011
15. Relative (ours):More natural than tallbuilding Less natural than forest More open than tallbuilding Less open than coastHas more perspective than tallbuildingBinary (existing):Not naturalNot openHas perspectiveUsing Relative Attributes15Parikh, Grauman, Relative Attributes, ICCV 2011
16. Relative (ours):More Young than CliveOwenLess Young than ScarlettJohanssonMore BushyEyebrows than ZacEfron Less BushyEyebrows than AlexRodriguezMore RoundFace than CliveOwenLess RoundFace than ZacEfronBinary (existing):Not YoungBushyEyebrowsRoundFaceUsing Relative Attributes16(Viggo)Parikh, Grauman, Relative Attributes, ICCV 2011