Visual Object Category Recognition

Total Page:16

File Type:pdf, Size:1020Kb

Visual Object Category Recognition Visual Object Category Recognition Robotics Research Group Department of Engineering Science University of Oxford Supervisors: Professor Andrew Zisserman Professor Pietro Perona Robert Fergus New College December 2, 2005 Abstract We investigate two generative probabilistic models for category-level object recognition. Both schemes are designed to learn categories with a minimum of supervision, requiring only a set of images known to contain the target category from a similar viewpoint. In both methods, learning is translation and scale-invariant; does not require alignment or correspondence between the training images, and is robust to clutter and occlusion. The schemes are also robust to heavy contamination of the training set with unrelated images, enabling them to learn directly from the output of Internet Image Search engines. In the first approach, category models are probabilistic constellations of parts, and their parameters are estimated by maximizing the likelihood of the training data. The appearance of the parts, as well as their mutual position, relative scale and probability of detection are explicitly represented. Recognition takes place in two stages. First, a feature-finder identifies promising locations for the model’s parts. Second, the category model is used to compare the likelihood that the observed features are generated by the category model, or are generated by background clutter. The second approach is a visual adaptation of “bag of words” models used to extract topics from a text corpus. We extend the approach to incorporate spatial information in a scale and translation-invariant manner. The model represents each image as a joint histogram of visual word occurrences and their locations, relative to a latent reference frame of the object(s). The model is entirely discrete, making no assumptions of uni-modality or the like. The parameters of the multi-component model are estimated in a maximum likelihood fashion over the training data. In recognition, the relative weighting of the different model components is computed along with the model reference frame with the highest likelihood, enabling the localization of object instances. The flexible nature of both models is shown by experiments on 28 datasets containing 12 diverse object categories, including geometrically constrained categories (e.g. faces, cars) and flexible objects (such as animals). The different datasets give a thorough evaluation of both methods in classification, categorization, localization and learning from contaminated data. This thesis is submitted to the Department of Engineering Science, University of Oxford, in fulfilment of the requirements for the degree of Doctor of Philosophy. This thesis is entirely my own work, and except where otherwise stated, describes my own research. Robert Fergus, New College Copyright c 2005 Robert Fergus All Rights Reserved To my parents Acknowledgements I would like to thank my two advisers: Professor Andrew Zisserman at Oxford and Professor Pietro Perona at Caltech for their guidance, patience and advice. This thesis has been a thrilling and rewarding experience thanks to them. I also thank Fei-Fei Li being my main collaborator over the last 5 years, both in classes and research. Many other people who have offered advice and guidance with my work for which I am very grateful (in alphabetical order): Andrew Blake, Mark Everingham, David Forsyth, Alex Holub, Dan Huttenlocher, Michael Isard, Jitendra Malik, Silvio Savarese, Josef Sivic, Frederik Schaffalitzky. I would also like to thank all the people in the Vision Labs at both Caltech and Oxford for making them such interesting places to be. I thank the various sources of funding I have had over the years: the Caltech CNSE, the UK EPSRC, EC project CogViSys and the PASCAL project. Agnes deserves special thanks for being so supportive of my efforts and so understanding of the endless deadlines. A final thanks must go to Pietro, Markus Weber and Max Welling who supervised me as a summer student while I was an undergraduate, sparking my interest in computer vision and object recognition. Contents Table of Contents i 1 Introduction 1 1.1Objective........................................ 1 1.2Motivation....................................... 2 1.3Challenges........................................ 4 1.4Definitionofvocabulary................................ 6 1.5 Contribution . ..................................... 7 1.5.1 TheConstellationModel........................... 7 1.5.2 Translation and Scale-Invariant pLSA (TSI-pLSA) ............. 8 1.6Outlineofthethesis.................................. 9 2 Literature review 10 2.1Specificinstancerecognition.............................. 11 2.1.1 Geometricmethods............................... 11 2.1.2 Globalappearancemethods.......................... 15 2.1.3 Texturedregionmethods........................... 16 2.2Categorylevelrecognition............................... 19 2.2.1 Digits...................................... 19 2.2.2 Faces,CarsandHumans............................ 21 2.2.3 Recent work . ................................. 25 2.2.4 Summaryofliterature............................. 35 2.3FeaturesandRepresentationSchemes........................ 36 2.3.1 Kadir&Brady................................. 36 2.3.2 Curves...................................... 37 2.3.3 Difference of Gaussians . ......................... 38 2.3.4 MultiscaleHarris................................ 38 2.3.5 SampledEdgeoperator............................ 39 2.3.6 Comparisonoffeaturedetectors....................... 41 2.3.7 SIFTdescriptor................................. 41 3 Datasets 43 3.1Caltechdatasets.................................... 44 3.2UIUCdataset...................................... 44 3.3FawltyTowers...................................... 45 3.3.1 TrainingdataforFawltyTowers....................... 45 3.4PASCALchallenge................................... 47 3.5Imagesearchenginedata............................... 48 3.6Summaryofdatasets.................................. 51 i 4 The Constellation model 53 4.1Introduction....................................... 53 4.2 Model inputs . ..................................... 54 4.3 Overview of model . ................................. 54 4.4Appearance....................................... 56 4.4.1 Appearancerepresentation.......................... 57 4.4.2 Curverepresentation.............................. 58 4.5Shape.......................................... 60 4.5.1 Fullmodel.................................... 60 4.5.2 Starmodel................................... 62 4.6Relativescale...................................... 63 4.7 Occlusion and Statistics of the feature finder . ................. 63 4.8Multipleaspectsviaamixtureofconstellationmodels............... 64 4.9Modeldiscussion.................................... 65 4.9.1 Appearanceterm................................ 65 4.9.2 Alternativeformsofshapemodel....................... 65 4.9.3 Improvements over Weber et al. ....................... 66 4.9.4 Modelassumptions............................... 67 4.10Modelstructuresummary............................... 67 5 Learning and Recognition with the Constellation model 69 5.1Learning......................................... 69 5.1.1 Initialization.................................. 71 5.1.2 EMupdateequations............................. 71 5.1.3 Computational considerations ......................... 73 5.1.4 Efficientsearchmethodsforthefullmodel.................. 74 5.1.5 Convergence.................................. 76 5.1.6 Backgroundmodel............................... 79 5.1.7 Finalmodel................................... 80 5.2Recognition....................................... 83 5.3Considerationsforthestarmodel........................... 84 5.3.1 Efficientmethodsforthestarmodel..................... 84 6 Weakly supervised experiments with the constellation model 89 6.1Fullmodelexperiments................................ 91 6.1.1 Baselineexperiments.............................. 96 6.2Analysisofperformance................................ 97 6.2.1 Changingscaleoffeatures........................... 98 6.2.2 FeatureRepresentation............................ 99 6.2.3 Numberofpartsinmodel...........................100 6.2.4 Contribution of the different model terms . .................101 6.2.5 Over-fitting . .................................103 6.2.6 Contaminationofthetrainingset.......................104 6.2.7 Samplingfromthemodel...........................105 6.3Comparisonwithothermethods...........................105 6.4Starmodelexperiments................................107 6.4.1 Comparisontofullmodel...........................107 6.4.2 Heterogeneouspartexperiments.......................107 6.4.3 Numberofpartsanddetections........................108 ii 7 Translation and Scale Invariant Probabilistic Latent Semantic Analysis 114 7.1 Probabilistic Latent Semantic Analysis (pLSA) . .................115 7.1.1 LatentDirichletAllocation(LDA)......................116 7.2ApplyingpLSAtovisualdata.............................117 7.2.1 Visualwords..................................117 7.2.2 Anexample...................................118 7.3AddinglocationintothepLSAmodel........................118 7.3.1 AbsolutePositionpLSA............................119 7.3.2 ScaleandTranslationInvariantpLSA....................119 7.4Regiondetectors....................................123 7.5Implementationaldetails................................124 7.6Caltechexperiments..................................125 7.7Modelinvestigations..................................132
Recommended publications
  • Learning and Using Taxonomies for Visual and Olfactory Category Recognition
    Learning and Using Taxonomies for Visual and Olfactory Category Recognition Thesis by Greg Griffin In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy California Institute of Technology Pasadena, California 2013 (Defended March 2013) ii c 2013 Greg Griffin All Rights Reserved iii For Leslie iv v Acknowledgements vi vii Abstract Dominant techniques for image classification over the last decade have mostly followed a three-step process. In the first step, local features (such as SIFT) are extracted. In the second step, some statistic is generated which compares the features in each new image to the features in a set of training images. Finally, standard learning models (such as SVMs) use these statistics to decide which class the image is most likely to belong to. Popular models such as Spatial Pyramid Matching and Naive Bayse Nearest Neighbors adhere to the same general linear 3-step process even while they differ in their choice of statistic and learning model. In this thesis I demonstrate a hierarhical approach which allows the above 3 steps to in- teract more efficiently. Just as a child playing “20 questions” seeks to identify the unknown object with the fewest number of querries, our learning model seeks to classify unlabelled data with a minimial set of feature querries. The more costly feature extraction is, the more beneficial this approach can be. We apply this approach to three different applications: photography, microscopy and olfaction. Image —. Microscope slide —. Finally, — both efficiency and accuracy. In our tests, the resulting system functions well under real-world outdoor conditions and can recognize a wide variety of real-world odors.
    [Show full text]
  • CS 558: Computer Vision 9Th Set of Notes
    1 CS 558: Computer Vision 9th Set of Notes Instructor: Philippos Mordohai Webpage: www.cs.stevens.edu/~mordohai E-mail: [email protected] Office: Lieb 215 Introduction to object recognition By Svetlana Lazebnik Slides adapted from Fei-Fei Li, Rob Fergus, Antonio Torralba, and Jean Ponce Overview • Basic recognition tasks • A machine learning approach • Example features • Example classifiers • Levels of supervision • Datasets • Current trends and advanced recognition tasks Specific recognition tasks Scene categorization • outdoor/indoor • city/forest/factory/etc. Image annotation/tagging • street • people • building • mountain • … Object detection • find pedestrians Activity recognition • walking • shopping • rolling a cart • sitting • talking • … Image parsing sky mountain building tree building banner street lamp market people Image understanding? How many visual object categories are there? Biederman 1987 OBJECTS ANIMALS PLANTS INANIMATE NATURAL MAN-MADE ….. VERTEBRATE MAMMALS BIRDS TAPIR BOAR GROUSE CAMERA Recognition: A machine learning approach Slides adapted from Fei-Fei Li, Rob Fergus, Antonio Torralba, Kristen Grauman, and Derek Hoiem The machine learning framework • Apply a prediction function to a feature representation of the image to get the desired output: f( ) = “apple” f( ) = “tomato” f( ) = “cow” The machine learning framework y = f(x) output prediction function Image feature • Training: given a training set of labeled examples {(x1,y1), …, (xN,yN)}, estimate the prediction function f by minimizing the prediction error on the training set • Testing: apply f to a never before seen test example x and output the predicted value y = f(x) Steps Training Training Labels Training Images Image Learned Training Features model Learned model Testing Image Prediction Features Test Image Slide credit: D.
    [Show full text]
  • Dataset Issues in Object Recognition
    Dataset Issues in Object Recognition J. Ponce1,2,T.L.Berg3, M. Everingham4,D.A.Forsyth1,M.Hebert5, S. Lazebnik1,M.Marszalek6,C.Schmid6, B.C. Russell7,A.Torralba7, C.K.I. Williams8, J. Zhang6, and A. Zisserman4 1 University of Illinois at Urbana-Champaign, USA 2 Ecole Normale Sup´erieure, Paris, France 3 University of California at Berkeley, USA 4 Oxford University, UK 5 Carnegie Mellon University, Pittsburgh, USA 6 INRIA Rhˆone-Alpes, Grenoble, France 7 MIT, Cambridge, USA 8 University of Edinburgh, Edinburgh, UK Abstract. Appropriate datasets are required at all stages of object recognition research, including learning visual models of object and scene categories, detecting and localizing instances of these models in im- ages, and evaluating the performance of recognition algorithms. Current datasets are lacking in several respects, and this paper discusses some of the lessons learned from existing efforts, as well as innovative ways to obtain very large and diverse annotated datasets. It also suggests a few criteria for gathering future datasets. 1 Introduction Image databases are an essential element of object recognition research. They are required for learning visual object models and for testing the performance of classification, detection, and localization algorithms. In fact, publicly available image collections such as UIUC [1], Caltech 4 [10], and Caltech 101 [9] have played a key role in the recent resurgence of category-level recognition research, driving the field by providing a common ground for algorithm development and evaluation.
    [Show full text]
  • Labelme: a Database and Web-Based Tool for Image Annotation Bryan C
    Computer Science and Artificial Intelligence Laboratory Technical Report MIT-CSAIL-TR-2005-056 September 8, 2005 AIM-2005-025 LabelMe: A Database and Web-based Tool for Image Annotation Bryan C. Russell, Antonio Torralba, Kevin P. Murphy, William T. Freeman massachusetts institute of technology, cambridge, ma 02139 usa — www.csail.mit.edu Abstract Research in object detection and recognition in cluttered scenes requires large image collections with ground truth labels. The labels should provide information about the object classes present in each image, as well as their shape and locations, and possibly other attributes such as pose. Such data is useful for testing, as well as for supervised learning. This project provides a web-based annotation tool that makes it easy to an- notate images, and to instantly share such annotations with the community. This tool, plus an initial set of 10,000 images (3000 of which have been labeled), can be found at http://www.csail.mit.edu/»brussell/research/LabelMe/intro.html1 1Financial support was provided by the National Geospatial-Intelligence Agency, NEGI-1582-04-0004, and a grant from BAE Systems. Kevin Murphy was supported in part by a Canadian NSERC Discovery Grant. 1 LabelMe: a database and web-based tool for image annotation BRYAN C. RUSSELL, ANTONIO TORRALBA, Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA 02139, USA [email protected], [email protected] KEVIN P. MURPHY Departments of computer science and statistics, University of British Columbia, Vancouver, BC V6T 1Z4 [email protected] WILLIAM T. FREEMAN Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA 02139, USA [email protected] Abstract Research in object detection and recognition in cluttered scenes requires large im- age collections with ground truth labels.
    [Show full text]
  • Labelme: a Database and Web-Based Tool for Image Annotation
    MIT AI LAB MEMO AIM-2005-025, SEPTEMBER, 2005. LabelMe: a database and web-based tool for image annotation BRYAN C. RUSSELL, ANTONIO TORRALBA, Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA 02139, USA [email protected], [email protected] KEVIN P. MURPHY Departments of computer science and statistics, University of British Columbia, Vancouver, BC V6T 1Z4 [email protected] WILLIAM T. FREEMAN Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA 02139, USA [email protected] Abstract Research in object detection and recognition in cluttered scenes requires large im- age collections with ground truth labels. The labels should provide information about the object classes present in each image, as well as their shape and locations, and possibly other attributes such as pose. Such data is useful for testing, as well as for supervised learning. This project provides a web-based annotation tool that makes it easy to annotate images, and to instantly share such annotations with the community. This tool, plus an initial set of 10,000 images (3000 of which have been labeled), can be found at http://www.csail.mit.edu/»brussell/research/LabelMe/intro.html 1 Introduction Detecting generic object categories in natural, cluttered images is one of the holy grails of computer vision. In order to make progress towards this goal, it may be essential to have large databases of challenging images, in which “ground truth” labels are made publically available. These labels should provide information about the object classes present in each image, as well as their shape and locations, and possibly other attributes 1 such as pose.
    [Show full text]
  • Topics of the Class
    Advanced Topics in Computer Vision" and Robotics " ! Jana Kosecka! http://cs.gmu.edu/~kosecka/cs884/! [email protected]! ! ! ! Some slides thanks to S. Lazebnik, T. Berg, Fei-Fei Li, K. Grauman and others! Topics of the class! •# State of the art of scene understanding! •# Object, Scene, Human Activity Recognition ! ! With applications to: ! ! •# Image Based Retrieval! •# Image tagging ! •# Robot Perception, environment understanding ! 1! Logistics! •# Grading: Homeworks, Presentations, Class Participation 60% Final exam/ project: 40% ! •# Prerequisites: Computer Vision, Robotics, AI, Data Mining, Pattern Recognition! •# Related Resources: Material covered in CS682, CS685 and textbooks and ! recommended materials there! •# Lectures: Introduction by an instructor, 3 paper presentations per class, discussions, each student will present one paper every second week; all students should read all papers to participate in discussion; programming homeworks every second week ! •# Projects: up to teams of 2 people! •# Dates! –# Project proposals due! –# May week of finals final report due! –# Project presentations! •# Required Software MATLAB (with Image Processing toolbox)! •# Open CV library! Student Participation, Presentation! •# 2-3 papers for each week discussing selected topic! •# 10-15 minute presentation of one paper with slides! •# Discuss the main idea of the paper! •# The methods used, if the code is available demonstrate the method! •# Provide opinion, compare related to the other papers on the same topic! •# Presenter should stimulate
    [Show full text]
  • LNCS 4170, Pp
    Dataset Issues in Object Recognition J. Ponce1,2,T.L.Berg3, M. Everingham4,D.A.Forsyth1,M.Hebert5, S. Lazebnik1,M.Marszalek6,C.Schmid6, B.C. Russell7,A.Torralba7, C.K.I. Williams8, J. Zhang6, and A. Zisserman4 1 University of Illinois at Urbana-Champaign, USA 2 Ecole Normale Sup´erieure, Paris, France 3 University of California at Berkeley, USA 4 Oxford University, UK 5 Carnegie Mellon University, Pittsburgh, USA 6 INRIA Rhˆone-Alpes, Grenoble, France 7 MIT, Cambridge, USA 8 University of Edinburgh, Edinburgh, UK Abstract. Appropriate datasets are required at all stages of object recognition research, including learning visual models of object and scene categories, detecting and localizing instances of these models in im- ages, and evaluating the performance of recognition algorithms. Current datasets are lacking in several respects, and this paper discusses some of the lessons learned from existing efforts, as well as innovative ways to obtain very large and diverse annotated datasets. It also suggests a few criteria for gathering future datasets. 1 Introduction Image databases are an essential element of object recognition research. They are required for learning visual object models and for testing the performance of classification, detection, and localization algorithms. In fact, publicly available image collections such as UIUC [1], Caltech 4 [10], and Caltech 101 [9] have played a key role in the recent resurgence of category-level recognition research, driving the field by providing a common ground for algorithm development and evaluation.
    [Show full text]