Download presentation
Presentation is loading. Please wait.
Published byChristine Short Modified over 9 years ago
1
ICG Professor Horst Cerjak, 19.12.2005 1 Horst Bischof Future Vision Future of Computer Vision Horst Bischof Inst. for Computer Graphics and Vision Graz University of Technology
2
ICG Professor Horst Cerjak, 19.12.2005 2 Horst Bischof Future Vision Motto of the talk It is a fantastic time …
3
ICG Professor Horst Cerjak, 19.12.2005 3 Horst Bischof Future Vision Motto of the talk to do computer vision!
4
ICG Professor Horst Cerjak, 19.12.2005 4 Horst Bischof Future Vision WHY?
5
ICG Professor Horst Cerjak, 19.12.2005 5 Horst Bischof Future Vision Computer Vision At least three goals Understand biological visual systems Build machines that see What are the fundamental processes of seeing
6
ICG Professor Horst Cerjak, 19.12.2005 6 Horst Bischof Future Vision Computer Vision The systems today are still exceedingly limited in their performance considerable room for improvement Where are chairs? Two interpretations? How many feet?
7
ICG Professor Horst Cerjak, 19.12.2005 7 Horst Bischof Future Vision Holy Grails in Vision 1.Segmentation 2. Correspondence Recognition Problem
8
ICG Professor Horst Cerjak, 19.12.2005 8 Horst Bischof Future Vision Future of Computer Vision Where do the innovations come from? 1. Hardware 2. Algorithms/Software
9
ICG Professor Horst Cerjak, 19.12.2005 9 Horst Bischof Future Vision HARDWARE
10
ICG Professor Horst Cerjak, 19.12.2005 10 Horst Bischof Future Vision Hardware First time that HW is no longer a real limitation !! Processing Image Resolution Storage Internet Mobile Devices Networks of cameras
11
ICG Professor Horst Cerjak, 19.12.2005 11 Horst Bischof Future Vision Processing Moore’s Law still holds! Multi-core CPUs Highly Parallel GPUs (+ Software eg. Cuda) DEMO
12
ICG Professor Horst Cerjak, 19.12.2005 12 Horst Bischof Future Vision Resolution 1900 Chicago & Alton Railroad Train (photograph a train), $5000 Ever growing resolution: 1975: 100 x 100 = 0.01 MP 2008: 9216 × 9216 = 85 MP (BAE) UltraCam x : 216 MP New fantastic opportunities Computational Cameras
13
ICG Professor Horst Cerjak, 19.12.2005 13 Horst Bischof Future Vision Internet Huge repository of images Flickr: 3.Nov. 2008 ~3 Billion Photos On-line 1 Million added a day YouTube: 65.000 new Videos a day 20% of Internet Traffic What can we do with these images?
14
ICG Professor Horst Cerjak, 19.12.2005 14 Horst Bischof Future Vision Mobile Vision Most of us have a mobile CV device with them Small Cameras Embedded Systems Mobile CV next large application area Place Recognition Recognizing Tags Shopping Games Augmented Reality 4,4mmx15mm
15
ICG Professor Horst Cerjak, 19.12.2005 15 Horst Bischof Future Vision ALGORITHMS
16
ICG Professor Horst Cerjak, 19.12.2005 16 Horst Bischof Future Vision (Some) New Developments Bayesian MethodsEnergy Minimization Discrete Continuous Machine Learning & Vision
17
ICG Professor Horst Cerjak, 19.12.2005 17 Horst Bischof Future Vision Bayesian Methods Lots of applications Computationally heavy Easily parallelizable Energy minimization approaches Ill-posed Prior Data
18
ICG Professor Horst Cerjak, 19.12.2005 18 Horst Bischof Future Vision Energy minimization Level SetsConvex formulationsGraph cuts Continuous Discrete Local optimaGlobal optima * GPU ImplementationMemory limitations Metrication errors Apapted f. D Cremers 2007 More to come
19
ICG Professor Horst Cerjak, 19.12.2005 19 Horst Bischof Future Vision Continous energy functional Data term potentially non-convex Global Optimal Solution Defines domain of application –Denoising –Segmentation –Stereo Total Variation regularization Data term Pock et.al
20
ICG Professor Horst Cerjak, 19.12.2005 20 Horst Bischof Future Vision Vision & Learning Combining Computer Vision with ML Huge Success We have good/stable features SIFT Boundary fragments If enough data learning works SVM Boosting
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.