Dudek & Jugessur, ICRA 2000. April 2000, IEEE ICRADudek & Jugessur Robust Place and Object Recognition using Local Appearance based Methods Gregory Dudek.

Slides:



Advertisements
Similar presentations
Tools of Composition in Photography. The photographers many choices… The photographers many choices… point of view focus placement background.
Advertisements

1 Benoit Boulet, Ph.D., Eng. Industrial Automation Lab McGill Centre for Intelligent Machines Department of Electrical and Computer Engineering McGill.
Towards Geographical Referencing of Monocular SLAM Reconstruction Using 3D City Models: Application to Real- Time Accurate Vision-Based Localization Reporter.
DONG XU, MEMBER, IEEE, AND SHIH-FU CHANG, FELLOW, IEEE Video Event Recognition Using Kernel Methods with Multilevel Temporal Alignment.
Object Recognition with Features Inspired by Visual Cortex T. Serre, L. Wolf, T. Poggio Presented by Andrew C. Gallagher Jan. 25, 2007.
Evaluating Color Descriptors for Object and Scene Recognition Koen E.A. van de Sande, Student Member, IEEE, Theo Gevers, Member, IEEE, and Cees G.M. Snoek,
Detection of Deviant Behavior From Agent Traces Boštjan Kaluža Department of Intelligent Systems, Jožef Stefan Institute Jozef Stefan Institute Jožef Stefan.
Face Alignment with Part-Based Modeling
Hybrid Position-Based Visual Servoing
A 4-WEEK PROJECT IN Active Shape and Appearance Models
Object retrieval with large vocabularies and fast spatial matching
RECOGNIZING FACIAL EXPRESSIONS THROUGH TRACKING Salih Burak Gokturk.
Face Recognition with Harr Transforms and SVMs EE645 Final Project May 11, 2005 J Stautzenberger.
4EyesFace-Realtime face detection, tracking, alignment and recognition Changbo Hu, Rogerio Feris and Matthew Turk.
Algebraic Functions of Views for 3D Object Recognition CS773C Advanced Machine Intelligence Applications Spring 2008: Object Recognition.
L ++ An Ensemble of Classifiers Approach for the Missing Feature Problem Using learn ++ IEEE Region 2 Student Paper Contest University of Maryland Eastern.
Illumination Normalization with Time-Dependent Intrinsic Images for Video Surveillance Yasuyuki Matsushita, Member, IEEE, Ko Nishino, Member, IEEE, Katsushi.
1 Invariant Local Feature for Object Recognition Presented by Wyman 2/05/2006.
Object recognition under varying illumination. Lighting changes objects appearance.
Geometrically overlaying di ff erent representations of an object in a scene By: Senate Taka CS 104 Final Project.
Visual Speech Recognition Using Hidden Markov Models Kofi A. Boakye CS280 Course Project.
Cultural Visual Literacy Images Americans Should Recognize.
Alternative Parallel Processing Approaches Jonathan Sagabaen.
Automated Patent Classification By Yu Hu. Class 706 Subclass 12.
CPS Graduate Seminar Student Presentations.
LECTURE 6 Segment-based Localization. Position Measurement Systems The problem of Mobile Robot Navigation: Where am I? Where am I going? How should I.
Video Tracking Using Learned Hierarchical Features
Camera/Vision for Geo-Location & Geo-Identification John S. Zelek Intelligent Human Machine Interface Lab Dept. of Systems Design Engineering University.
Yao, B., and Fei-fei, L. IEEE Transactions on PAMI(2012)
Incremental learning for Robust Visual Tracking Ko Dae-Won.
Self-Calibration and Metric Reconstruction from Single Images Ruisheng Wang Frank P. Ferrie Centre for Intelligent Machines, McGill University.
D31 Entity Recognition Results with Auto- associative Memories Nicolas Gourier INRIA PRIMA Team GRAVIR Laboratory CAVIAR Project.
University of Amsterdam Search, Navigate, and Actuate - Qualitative Navigation Arnoud Visser 1 Search, Navigate, and Actuate Qualitative Navigation.
Deformable Part Model Presenter : Liu Changyu Advisor : Prof. Alex Hauptmann Interest : Multimedia Analysis April 11 st, 2013.
Navigating 3D Worlds via 2D Multi- Touch Interfaces Daniel Cope Supervised by Stuart Marshall 1.
Objectives Objectives Recommendz: A Multi-feature Recommendation System Matthew Garden, Gregory Dudek, Center for Intelligent Machines, McGill University.
NTIT IMD 1 Speaker: Ching-Hao Lai( 賴璟皓 ) Author: Hongliang Bai, Junmin Zhu and Changping Liu Source: Proceedings of IEEE on Intelligent Transportation.
Image Registration with Hierarchical B-Splines Z. Xie and G. Farin.
Design of PCA and SVM based face recognition system for intelligent robots Department of Electrical Engineering, Southern Taiwan University, Tainan County,
Based on the success of image extraction/interpretation technology and advances in control theory, more recent research has focused on the use of a monocular.
COMP 417 – Jan 12 th, 2006 Guest Lecturer: David Meger Topic: Camera Networks for Robot Localization.
Pure Topological Mapping in Mobile Robotics Authors : Dimitri Marinakis Gregory Dudek Speaker :李宗明 M99G0103 IEEE TRANSACTIONS ON ROBOTICS, VOL. 26, NO.
Face detection and recognition Many slides adapted from K. Grauman and D. Lowe.
Application of (Multi-layer) Perceptron Areas of application include: Speech recognition Image recognition Machine translation Cyber security.
Image from
Biologically Inspired Vision-based Indoor Localization Zhihao Li, Ming Yang
Age-invariant Face Recognition
Face Recognition based on 2D-PCA and CNN
3D Visual Phrases for Landmark Recognition
COGNITIVE APPROACH TO ROBOT SPATIAL MAPPING
Can Computer Algorithms Guess Your Age and Gender?
Application of (Multi-layer) Perceptron
Sample Title First Name, Last Name, Degree
Picode: A New Picture-Embedding 2D Barcode
Final Year Project Presentation --- Magic Paint Face
Absolute & Relative Location & GPS Technology
Object Recognition in the Dynamic Link Architecture
Context-based vision system for place and object recognition
Self transforming to power law topology for overlay networks.
Tools of Composition in Photography.
Geometric Hashing: An Overview
Aim of the project Take your image Submit it to the search engine
George Bebis and Wenjing Li Computer Vision Laboratory
Paper Reading Dalong Du April.08, 2011.
Outline A. M. Martinez and A. C. Kak, “PCA versus LDA,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 23, no. 2, pp , 2001.
Around the room Orders of operations.
AHED Automatic Human Emotion Detection
Digital image Levels of gray levels, quality: 1 byte = 8 bit 0 = Black
A. Author1, B. Author2 and C. Author3
Onboard Flight Management with Artificial Intelligence
Presentation transcript:

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Robust Place and Object Recognition using Local Appearance based Methods Gregory Dudek and Deeptiman Jugessur Center for Intelligent Machines McGill University +

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Outline Applications PCA: shortcomings Objectives Approach Background System Overview Results Conclusion

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Two Applications Object recognition: what is that thing? –Recognizing a known object from its visual appearance. –Landmarks, grasping targets, etc. Place recognition (coarse localization): what room am I in? –Recognizing the current waypoint on a trajectory, validating the current locale for the application of a precise localization method, topological navigation.

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur PCA-based recognition. Has now become a well established method for image recognition. PCA-based recognition: global transform of image with N degrees of freedom into an eigenspace with M << N degrees of freedom. –Freedoms M are the “most important” characteristics of the set of images being memorized. Avoids having to segment image into object & background by using the whole thing.

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Observations Using whole image implies recognizing combination of object AND background. Segmenting object from background would avoid dependence on background, but it’s too difficult. Using a small sub-region gives a less precise recognition (e.e. the sun-window could come from more than one image), it’s is efficient. Many subwindows together can “vote” for an unambiguous recognition. If the sub-windows are suitably chosen, they may totally ignore the background.

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Problem Statement Improving the performance of classic PCA based recognition by accounting for: –Varying backgrounds –Planar rotations –Occlusions Also (discussed in less detail) –Changes in object pose –Non-rigid deformation

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Our key idea(s). Use sub-windows: several together uniquely accomplish recognition. Sub-windows are selected by an attention operator (several kinds can be used). Each sub-window is sampled non-uniformly to weight it towards it’s center. Use only the amplitude spectrum to buy rotational invariance.

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Background Standard Appearance Based Recognition –M. Turk and S. Pentland 1991 –S.K. Nayar, H. Murase, S.A. Nene 1994 –H. Murase, S.K. Nayar 1995 –Shortcomings (due to global approach): Background Scale Rotations Local changes of the image or object Occlusion

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Background (part 2) “Enhanced” Local sub-window methods –D. Lowe 1999: scale invariance, simple features. –C. Schmid 1999: Probabilistic approach based on sub-windows extracted using Harris operator. –C. Schmid & R. Mohr 1997: numerous sub-windows extracted using Harris operator for database image retrieval (simpler problem). –K. Ohba & K. Ikeuchi 1997: K.L.T. operator used for the extraction of sub-windows for the creation of an eigenspace. Only handles occlusion. Interest Operator of choice: –D. Reisfeld, H. Wolfson, Y.Yeshurun 1995: Local symmetry operator

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Approach 2 phases: –Training (off-line) for the entire database of recognizable images: Run an interest operator to obtain a saliency map for each image. Choose sub-windows around the salient points for each image. Select most informative sub-windows and use foveal sampling. Create the eigenspace with the processed sub-windows. –Testing (on-line) for a candidate test image: Run the same interest operator to obtain the saliency map. Choose the sub-windows and process the information within them. Project the sub-windows onto the eigenspace Perform classification based on nearest neighbor rules.

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Recognition Model Database of recognizable images Candidate test image Extract sub-windows based on interest operator saliency values and information content Obtain amplitude spectra for the sub-windows Eigenspace for classification Run all images though the interest operator Run the image through the interest operator 2D FFT Create low dim. eigenspace Project onto eigenspace Off-line On-line

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Polar Samplings and 2D FFT Polar Sampling Same Amplitude Spectrum (in theory) 2D FFT

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Shift Theorem

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Place Recognition Test ImagesTraining Images Best match

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Place Recognition (2) Test Images Training Images Best match

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Object Recognition Test Image Training Image Recognition

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Object Recognition (2) Test ImageTraining Image Best matches Note: background variation and occlusion

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Performance metrics On-line performance: 15x15 pixel subwindows: 90% recognition with 10 subwindows (10 interest points). 15x15 pixel subwindows: 100% recognition using 15 more subwindows –Interest operator can take 1/30s to 10 min. (depending on the operator, images size, etc.). –Classification in Eigenspace well under 1 sec (can be performed in real time).

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Performance vs Number of Interest Points Recognition Rate 100% Number of features Note: 10 windows of size 15x15 means using only 0.7% of the total image content.

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Conclusion & Extensions Approach to object and place recognition from single video images. Works despite planar rotation, occlusion or other deformations. Highly robust. Recognition rates of up to 100% with 20 test images. Improved robustness to background can be achieved using “masking” [Jugessur & Dudek CVPR 2000]. Ongoing work sees to exploit geometry of interest points. Could filter in Eigenspace during training to select only “useful” features.

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur That’s all

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Questions you could ask Have you considered the use of alternative interest/attention operators? Does the operator matter? What if the background is much more interesting (to the operator) that the object? How much does color information matter? What is the consequence of not using geometric information (and what does that really mean)?

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Robust Place and Object Recognition using Local Appearance based Methods Gregory Dudek and Deeptiman Jugessur Center for Intelligent Machines McGill University

Dudek & Jugessur, ICRA April 2000, IEEE ICRADudek & Jugessur Performance metrics Training time: roughly 64 windows, 15x15, 17 objects, 3 views per object: 24 hours. –This is using MATLAB and highly non-optimized code. Using similar methods on global images, other groups have reported times on the order of minutes for similar tasks. On-line performance: –Interest operator can take 1/30s to 10 min. (depending on the operator, images size, etc.) –Classification in Eigenspace well under 1 sec (can be performed in real time).