1.13k likes | 1.5k Views
Attention in Computer Vision. Mica Arie-Nachimson and Michal Kiwkowitz May 22, 2005 Advanced Topics in Computer Vision Weizmann Institute of Science. Problem definition – Search Order. Vision applications apply “expensive” algorithms (e.g. recognition) to image patches
E N D
Attention in Computer Vision Mica Arie-Nachimson and Michal Kiwkowitz May 22, 2005 Advanced Topics in Computer Vision Weizmann Institute of Science
Problem definition – Search Order • Vision applications apply “expensive” algorithms (e.g. recognition) to image patches • Mostly naïve selection of patches • Selection of patches determines number of calls to “expensive” algorithm Object recognition NO
Problem Definition - Search Order • More sophisticated selection of patches would imply less calls to “expensive” algorithm • Attention used to efficiently focus on incoming data (better use for limited processing capacity) Object recognition YES NO
5 3 2 4 1 6 Problem Definition - Search Order Object recognition
Outline • What is Attention • Attention in Object Recognition • Saliency Model • Feature Integration Theory • Saliency Algorithm • Saliency & Object Recognition • Comparison • Inner Scene Similarity Model • Biological motivation • Difficulty of Search Tasks • Algorithms • FLNN • VSLE
Outline • What is Attention • Attention in Object Recognition • Saliency Model • Feature Integration Theory • Saliency Algorithm • Saliency & Object Recognition • Comparison • Inner Scene Similarity Model • Biological motivation • Difficulty of Search Tasks • Algorithms • FLNN • VSLE
Attention • Attention implies allocating resources, perceptual or cognitive, to some things at the expense of not allocating them to something else.
What is Attention • You are sitting in class listening to a lecture. • Two people behind you are talking. • Can you hear the lecture? • One of them mentions the name of a friend of yours. • How did you know?
Attention in Other Applications • Face Detection (feature selection) • Video Analysis (temporal block selection) • Robot Navigation (select locations) • …
Attention is Directed by: • Bottom-up: • From small to large units of meaning • Rapid • Task-independent
http://www.rybak-et-al.net/nisms.html Attention is Directed by: • Top-down: • Use higher levels (context, expectation) to process incoming information (Guess) • Slower • Task dependent
Outline • What is Attention • Attention in Object Recognition • Saliency Model • Feature Integration Theory • Saliency Algorithm • Saliency & Object Recognition • Comparison • Inner Scene Similarity Model • Biological motivation • Difficulty of Search Tasks • Algorithms • FLNN • VSLE
WHICH? Attention When is information selected (filtered)? • Early selection (Broadbent, 1958) • Cocktail party phenomenon (Moray, 1959) • Late selection (Treisman, 1960) - attenuation • All information is sent to perceptual systems for processing • Some is selected for complete processing • Some is more likely to be selected
Parallel Search Is there a green O ? + A. Treisman, G. Gelade, 1980
Conjunction Search Is there a green N ? + A. Treisman, G. Gelade, 1980
Results A. Treisman, G. Gelade, 1980
Conjunction Search + A. Treisman, G. Gelade, 1980
Color map Orientation map A. Treisman, G. Gelade, 1980
Color map Orientation map A. Treisman, G. Gelade, 1980
Conjunction Search + A. Treisman, G. Gelade, 1980
Intensity Curvature P P P P P P P P P P Line End Orientation P P x x x P P P s P x I x I I P P I I Color x x x x x Primitives Movement x
Feature Integration Theory Attention - two stages: • Attention • Serial Processing • Localized Focus • Slower • Conjunctive search • Pre-attention • Parallel Processing • Low Level Features • Fast • Parallel Search How is the Focus found & shifted? A. Treisman, G. Gelade, 1980
Outline • What is Attention • Attention in Object Recognition • Saliency Model • Feature Integration Theory • Saliency Algorithm • Saliency & Object Recognition • Comparison • Inner Scene Similarity Model • Biological motivation • Difficulty of Search Tasks • Algorithms • FLNN • VSLE
Attention Shifts in Attention “Shifts in selective visual attention: towards the underlying neural circuitry”, Christof Koch, and Shimon Ullman, 1985 • Feature • Maps • Orientation • Color • Curvature • Line end • Movement • Feature • Maps • Orientation • Color • Curvature • Line end • Movement • Feature • Maps • Orientation • Color • Curvature • Line end • Movement • Feature • Maps • Orientation • Color • Curvature • Line end • Movement Saliency • Feature • Maps • Orientation • Color • Curvature • Line end • Movement Central Representation C. Koch, and S. Ullman, 1985
Saliency • Salient - stands out “A model of saliency-based visual attention for rapid scene analysis” Laurent Itti, Christof Koch, and Ernst Niebur, 1998 • Example – telephone & road sign have high saliency L. Itti, C. Koch, and E. Niebur, 1998
from C. Koch L. Itti, C. Koch, and E. Niebur, 1998
Intensity Cells in the retina L. Itti, C. Koch, and E. Niebur, 1998
0 1 2 8 Intensity Create 8 spatial scale using Gaussian pyramids L. Itti, C. Koch, and E. Niebur, 1998
- + + Fine scale - Coarse scale Intensity Center-Surround difference operator • Sensitive to local spatial discontinuities • Principle computation in the retina & primary visual cortex • Subtract coarse scale from fine scale fine coarse L. Itti, C. Koch, and E. Niebur, 1998
Point-by-point subtraction Gauss Pyramid Interpolation Toy Example Fine level Coarse level Coarse level
Point-by-point subtraction Gauss Pyramid Interpolation Toy Example Fine level Coarse level Coarse level
Intensity Different ratios – multiscale feature extraction Compute: 6 Intensity maps L. Itti, C. Koch, and E. Niebur, 1998
Color Kandel et al. (2000). Principles of Neural Science. McGraw-Hill/Appleton & Lange Same c and s as with intensity 12 Color maps L. Itti, C. Koch, and E. Niebur, 1998 More
Same c and s as with intensity 24 Orientation maps From Visual system presentation by S. Ullman Orientation L. Itti, C. Koch, and E. Niebur, 1998 More
from C. Koch L. Itti, C. Koch, and E. Niebur, 1998
Normalization Operator More L. Itti, C. Koch, and E. Niebur, 1998
Saliency Map L. Itti, C. Koch, and E. Niebur, 1998
Algorithm- up to now 1. Extract Feature Maps 2. Compute Center-Surround (42) • Intensity – I (6) • Color – C (12) • Orientation – O (24) 3. Combine each channel into conspicuity map 4. Compute saliency by summing and normalizing maps
Winner Takes All Selection (FOA) Leaky integrate-and-fire neurons “Inhibition of return” FOA – Focus Of Attention L. Itti, C. Koch, and E. Niebur, 1998
Inhibition of return ends Results • FOA shifts: 30-70 ms • Inhibition: 500-900 ms L. Itti, C. Koch, and E. Niebur, 1998
Results Image Saliency SFC Output Spatial Frequency Content, Reinage & Zador, 1997 L. Itti, C. Koch, and E. Niebur, 1998
Results Image (a) (b) Saliency (c) SFC (d) Output Spatial Frequency Content, Reinage & Zador, 1997 L. Itti, C. Koch, and E. Niebur, 1998
Outline • What is Attention • Attention in Object Recognition • Saliency Model • Feature Integration Theory • Saliency Algorithm • Saliency & Object Recognition • Comparison • Inner Scene Similarity Model • Biological motivation • Difficulty of Search Tasks • Algorithms • FLNN • VSLE
Attention & Object Recognition • “Is bottom-up attention useful for object recognition?” • U. Rutishauser, D. Walther, C. Koch and P. Perona, 2004 Attention U. Rutishauser, D. Walther, C. Koch and P. Perona, 2004
saliency model Growing region in strongest map To Object Recognition (Lowe) Object Recognition U. Rutishauser, D. Walther, C. Koch and P. Perona, 2004 More
Attention & Object Recognition Learning inventories – “grocery cart problem” Real world scenes 1 image for training (15 fixations) 2-5 images for testing (20 fixations) U. Rutishauser, D. Walther, C. Koch and P. Perona, 2004
testing training Object recognition Match
“Grocery Cart” Problem training testing1 testing2 U. Rutishauser, D. Walther, C. Koch and P. Perona, 2004