Slide 1Slide 2Why is Vision Interesting?Vision is inferential: LightVision is InferentialSlide 6Computer VisionA Quick Tour of Computer VisionBoundary DetectionSlide 10Slide 11TrackingSlide 13Slide 14Slide 15Slide 16StereoSlide 18MotionMotion - ApplicationPose DeterminationRecognition - ShadingSlide 23ClassificationVision depends on:Slide 26Modeling + AlgorithmsBayesian inferenceEngineeringThe State of Computer VisionSlide 32Related FieldsComputer VisionJanuary 2004 L1.1© 2004 by Davi GeigerIntroductionComputer VisionJanuary 2004 L1.2© 2004 by Davi GeigerVision•``to know what is where, by looking.’’ (Marr).•Where•WhatComputer VisionJanuary 2004 L1.3© 2004 by Davi GeigerWhy is Vision Interesting?•Psychology–~ 35% of cerebral cortex is for vision.–Vision is how we experience the world.•Engineering–Want machines to interact with world.–Digital images are everywhere.Computer VisionJanuary 2004 L1.4© 2004 by Davi GeigerVision is inferential: Light(http://www-bcs.mit.edu/people/adelson/checkershadow_illusion.html)Computer VisionJanuary 2004 L1.5© 2004 by Davi GeigerVision is InferentialComputer VisionJanuary 2004 L1.6© 2004 by Davi GeigerVision is Inferential: Prior KnowledgeComputer VisionJanuary 2004 L1.7© 2004 by Davi GeigerComputer Vision•Inference  Computation•Building machines that see•Modeling biological perceptionComputer VisionJanuary 2004 L1.8© 2004 by Davi GeigerA Quick Tour of Computer VisionComputer VisionJanuary 2004 L1.9© 2004 by Davi GeigerBoundary Detectionhttp://www.robots.ox.ac.uk/~vdg/dynamics.htmlComputer VisionJanuary 2004 L1.10© 2004 by Davi GeigerComputer VisionJanuary 2004 L1.11© 2004 by Davi GeigerBoundary DetectionFinding the Corpus Callosum (G. Hamarneh, T. McInerney, D. Terzopoulos)Computer VisionJanuary 2004 L1.12© 2004 by Davi GeigerTrackingComputer VisionJanuary 2004 L1.13© 2004 by Davi GeigerTrackingComputer VisionJanuary 2004 L1.14© 2004 by Davi GeigerTrackingComputer VisionJanuary 2004 L1.15© 2004 by Davi GeigerTrackingComputer VisionJanuary 2004 L1.16© 2004 by Davi GeigerTrackingComputer VisionJanuary 2004 L1.17© 2004 by Davi GeigerStereoComputer VisionJanuary 2004 L1.18© 2004 by Davi GeigerStereohttp://www.magiceye.com/Computer VisionJanuary 2004 L1.19© 2004 by Davi GeigerMotionhttp://www.ai.mit.edu/courses/6.801/lect/lect01_darrell.pdfComputer VisionJanuary 2004 L1.20© 2004 by Davi GeigerMotion - Application(www.realviz.com)Computer VisionJanuary 2004 L1.21© 2004 by Davi GeigerPose DeterminationVisually guided surgeryComputer VisionJanuary 2004 L1.22© 2004 by Davi GeigerRecognition - ShadingLighting affects appearanceComputer VisionJanuary 2004 L1.23© 2004 by Davi GeigerComputer VisionJanuary 2004 L1.24© 2004 by Davi GeigerClassification(Funkhauser, Min, Kazhdan, Chen, Halderman, Dobkin, Jacobs)Computer VisionJanuary 2004 L1.25© 2004 by Davi GeigerVision depends on:•Geometry•Physics•The nature of objects in the world (This is the hardest part).Computer VisionJanuary 2004 L1.26© 2004 by Davi GeigerApproaches to VisionComputer VisionJanuary 2004 L1.27© 2004 by Davi GeigerModeling + Algorithms•Build a simple model of the world (eg., flat, uniform intensity).•Find provably good algorithms.•Experiment on real world.•Update model.Problem: Too often models are simplistic or intractable.Computer VisionJanuary 2004 L1.28© 2004 by Davi GeigerBayesian inference•Bayes law: P(A|B) = P(B|A)*P(A)/P(B).•P(world|image) = P(image|world)*P(world)/P(image)•P(image|world) is computer graphics–Geometry of projection.–Physics of light and reflection.•P(world) means modeling objects in world. Leads to statistical/learning approaches.Problem: Too often probabilities can’t be known and are invented.Computer VisionJanuary 2004 L1.29© 2004 by Davi GeigerEngineering•Focus on definite tasks with clear requirements.•Try ideas based on theory and get experience about what works.•Try to build reusable modules.Problem: Solutions that work under specific conditions may not generalize.Computer VisionJanuary 2004 L1.31© 2004 by Davi GeigerThe State of Computer Vision•Science–Study of intelligence seems to be hard.–Some interesting fundamental theory about specific problems.–Limited insight into how these interact.Computer VisionJanuary 2004 L1.32© 2004 by Davi GeigerThe State of Computer Vision•Technology–Interesting applications: inspection, graphics, security, internet….–Some successful companies. Largest ~100-200 million in revenues. Many in-house applications.–Future: growth in digital images exciting.Computer VisionJanuary 2004 L1.33© 2004 by Davi GeigerRelated Fields•Graphics. “Vision is inverse graphics”.•Visual perception.•Neuroscience.•AI•Learning•Math: eg., geometry, stochastic