modern features-part-2-descriptors
TRANSCRIPT
![Page 1: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/1.jpg)
Descriptors
Cordelia Schmid Tinne Tuytelaars
Krystian Mikolajczyk
![Page 2: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/2.jpg)
Overview – descriptors
• Introduction
• Modern descriptors
• Comparison and evaluation
![Page 3: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/3.jpg)
Descriptors
Extract affine regions Normalize regions Eliminate rotational
+ illumination Compute appearance
descriptors
SIFT (Lowe ’04)
![Page 4: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/4.jpg)
Descriptors - history • Normalized cross-correlation (NCC) [~ 60s]
• Gaussian derivative-based descriptors – Differential invariants [Koenderink and van Doorn’87] – Steerable filters [Freeman and Adelson’91]
• Moment invariants [Van Gool et al.’96]
• SIFT [Lowe’99]
• Shape context [Belongie et al.’02]
• Gradient PCA [Ke and Sukthankar’04]
• SURF descriptor [Bay et al.’08]
• DAISY descriptor [Tola et al.’08, Windler et al’09]
• …….
![Page 5: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/5.jpg)
SIFT descriptor [Lowe’99]
• Spatial binning and binning of the gradient orientation • 4x4 spatial grid, 8 orientations of the gradient, dim 128 • Soft-assignment to spatial bins • Normalization of the descriptor to norm one (robust to illumination) • Comparison with Euclidean distance
gradient
→ →
image patch
y
x
![Page 6: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/6.jpg)
Local descriptors - rotation invariance
• Estimation of the dominant orientation – extract gradient orientation – histogram over gradient orientation – peak in this histogram
• Rotate patch in dominant direction
0 2 π
The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still appears, you may have to delete the image and then insert it again.
• Plus: invariance • Minus: less discriminant, additional noise
![Page 7: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/7.jpg)
Evaluation • Descriptors should be
– Distinctive (! importance of the measurement region) – Robust to changes on viewing conditions as well as to errors of
the detector – Compactness, speed of computation
• Detection rate (recall) – #correct matches / #correspondences
• False positive rate – #false matches / #all matches
• Variation of the distance threshold – distance (d1, d2) < threshold
1
1
[K. Mikolajczyk & C. Schmid, PAMI’05]
![Page 8: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/8.jpg)
Viewpoint change (60 degrees)
* * Detector: Hessian-Affine
![Page 9: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/9.jpg)
* *
Scale change (factor 2.8) Detector: Hessian-Affine
![Page 10: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/10.jpg)
Evaluation - conclusion
• SIFT based descriptors perform best
• Significant difference between SIFT and low dimension descriptors as well as cross-correlation
• Robust region descriptors better than point-wise descriptors
• Performance of the descriptor is relatively independent of the detector
![Page 11: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/11.jpg)
Recent extensions to SIFT
• Color SIFT [Sande et al. 2010]
• Normalizing SIFT with square root transformation
[Arandjelovic et Zisserman’12]
![Page 12: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/12.jpg)
Descriptors – dense extraction
• Many conclusions for descriptors applied to sparse detectors also hold for dense extraction – For example normalizing SIFT with the square root improves in
image retrieval and classification
• Image retrieval: sparse versus dense
SIFT + Fisher vector for retrieval on the Holidays dataset
Hessian-Affine MAP = 0.54 Dense MAP = 0.62
![Page 13: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/13.jpg)
Overview – descriptors
• Introduction
• Modern descriptors
• Comparison and evaluation
![Page 14: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/14.jpg)
Overview
• Introduction
• Modern descriptors
• Comparison and evaluation
![Page 15: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/15.jpg)
Modern descriptors
• Efficient descriptors • Compact binary descriptors • More robust descriptors • Learned descriptors
![Page 16: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/16.jpg)
DAISY
• Op<mized for dense sampling • Log-‐polar grid • Gaussian smoothing • Dealing with occlusions
Engin Tola, Vincent Lepe<t, and Pascal Fua, DAISY: An Efficient Dense Descriptor Applied to Wide-‐Baseline Stereo, TPAMI 32(5), 2010.
Cita%ons: 150 (2012)
![Page 17: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/17.jpg)
SURF: Speeded Up Robust Features
• Approximate deriva<ves with Haar wavelets • Exploit integral images
Herbert Bay, Andreas Ess, Tinne Tuytelaars, Luc Van Gool "SURF: Speeded Up Robust Features", Computer Vision and Image Understanding (CVIU), Vol. 110, No. 3, pp. 346-‐-‐359, 2008
Cita%ons: 4500 (2012)
![Page 18: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/18.jpg)
SURF: Speeded Up Robust Features
• Orienta<on assignment
![Page 19: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/19.jpg)
U-‐SURF: Upright SURF
• Rota<on invariance is o`en not needed.
Don’t use more invariance than needed for a given applica<on
• Orienta<on es<ma<on takes <me. • Orienta<on es<ma<on is o`en a source of errors.
![Page 20: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/20.jpg)
Beyond the classics
• Efficient descriptors • Compact binary descriptors • More robust descriptors • Learned descriptors
![Page 21: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/21.jpg)
Fast and compact descriptors
• (Very) large scale applica<ons – > memory issues, computa<on <me issues
• Mobile phone applica<ons
![Page 22: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/22.jpg)
Fast and compact descriptors
• Binary descriptors • Comparison of pairs of intensity values -‐ LBP -‐ BRIEF -‐ ORB -‐ BRISK
![Page 23: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/23.jpg)
LBP: Local Binary Paeerns
T. Ojala, M. Pie<käinen, and D. Harwood (1994), "Performance evalua<on of texture measures with classifica<on based on Kullback discrimina<on of distribu<ons", ICPR 1994, pp.582-‐585. M Heikkilä, M Pie<käinen, C Schmid, Descrip<on of interest regions with LBP, Paeern recogni<on 42 (3), 425-‐436
• First proposed for texture recogni<on in 1994.
Cita%ons: 2500 (2012)
![Page 24: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/24.jpg)
BRIEF: Binary Robust Independent
Elementary Features • Random selec<on of pairs of intensity values.
• Fixed sampling paeern of 128, 256 or 512 pairs.
• Hamming distance to compare descriptors (XOR).
M. Calonder, V. Lepe<t, C. Strecha, P. Fua, BRIEF: Binary Robust Independent Elementary Features, 11th European Conference on Computer Vision, 2010.
Cita%ons: 149 (2012)
![Page 25: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/25.jpg)
D-‐BRIEF: Discrimina<ve BRIEF
T. Trzcinski and V. Lepe<t, Efficient Discrimina<ve Projec<ons for Compact Binary Descriptors European Conference on Computer Vision (ECCV) 2012
• Learn linear projec<ons that map image patches to a more discrimina<ve subspace
• Exploit integral images
![Page 26: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/26.jpg)
ORB: Oriented FAST and Rotated BRIEF
• Add rota<on invariance to BRIEF • Orienta<on assignment based on the intensity centroid
Ethan Rublee, Vincent Rabaud, Kurt Konolige, Gary Bradski, ORB: an efficient alterna<ve to SIFT or SURF, ICCV 2011
Cita%ons: 43 (2012)
![Page 27: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/27.jpg)
• Select a good set of pairwise comparisons: minimize correla<on under various orienta<on changes
ORB: Oriented FAST and Rotated BRIEF
High variance High variance + uncorrelated
![Page 28: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/28.jpg)
BRISK: Binary Robust Invariant Scalable
Keypoints • Regular grid
• Orienta<on assignment based on dominant gradient direc<on (using long-‐distance pairs)
• 512 bit descriptor based on short-‐distance pairs
Stefan Leutenegger, Margarita Chli and Roland Y. Siegwart, BRISK: Binary Robust Invariant Scalable Keypoints, ICCV 2011
Cita%ons: 20 (2012)
![Page 29: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/29.jpg)
Various others
• FREAK: Fast Re<na Keypoint • CARD: Compact and Real<me Descriptor • LDB: Local Difference Binary
![Page 30: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/30.jpg)
FREAK: Fast Re<na Keypoint
A. Alahi, R. Or<z, and P. Vandergheynst. FREAK: Fast Re<na Keypoint. In IEEE Conference on Computer Vision and Paeern Recogni<on 2012.
• Inspired by the human visual system
![Page 31: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/31.jpg)
CARD: Compact and Real<me Descriptor
– Look up tables – learning-‐based sparse hashing
![Page 32: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/32.jpg)
LDB: Local Difference Binary
Xin Yang and Kwang-‐Ting Cheng, LDB: An Ultra-‐Fast Feature for Scalable Augmened Reality on Mobile Devices, Interna<onal Symposium on Mixed and Augmented Reality 2012 (ISMAR 2012)
![Page 33: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/33.jpg)
Beyond the classics
• Efficient descriptors • Compact binary descriptors • More robust descriptors • Learned descriptors
![Page 34: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/34.jpg)
LIOP: Local Intensity Order Paeern for
Feature Descrip<on
Zhenhua Wang Bin Fan Fuchao Wu, Local Intensity Order Paeern for Feature Descrip<on, ICCV 2011.
• Robustness to monotonic intensity changes • Data-‐driven division into cells
(and predecessors MROGH and MRRID)
![Page 35: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/35.jpg)
LIOP: Local Intensity Order Paeern for
Feature Descrip<on
![Page 36: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/36.jpg)
Beyond the classics
• Efficient descriptors • Compact binary descriptors • More robust descriptors • Learned descriptors
![Page 37: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/37.jpg)
Winder & Brown
M. Brown, G. Hua and S. Winder, Discriminant Learning of Local Image Descriptors.. IEEE Transac<ons on Paeern Analysis and Machine Intelligence. 2010.
• Learn configura<on and other parameters from training data obtained from 3D reconstruc<ons
Cita%ons: 194 (2012)
![Page 38: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/38.jpg)
Winder & Brown
• Training data = set of corresponding image patches
![Page 39: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/39.jpg)
Descriptor Learning Using Convex Op<misa<on
• Convex learning of – spa<al pooling regions – dimensionality reduc<on
• Learning from very weak supervision
Non-‐linear transform
Spa<al pooling
Dimensionality reduc<on
Pre-‐rec<fied keypoint patch
Descriptor vector
Normalisa<on and cropping
learning
learning
K. Simonyan et al., Descriptor Learning Using Convex Op<misa<on, ECCV 2012
![Page 40: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/40.jpg)
Learning Spa<al Pooling Regions
• Selec%on from a large pool using L1 regularisa%on • So`-‐margin constraints:
– squared L2 distance between descriptors of matching feature pairs should be smaller than that of non-‐matching pairs
• Convex objec<ve (op<mised with a proximal method)
pooling region configura%ons learnt with different levels of sparsity 768-‐D 576-‐D 320-‐D
![Page 41: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/41.jpg)
Overview
• Introduction
• Modern descriptors
• Comparison and evaluation
![Page 42: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/42.jpg)
Sepng up an evalua<on • Which problem? Performance in different applica<on/niches may
vary significantly. – Category recogni<on, – Matching, – Retrieval
• What dataset? – Pascal VOC 2007 – Oxford image pairs – Oxford -‐ Paris buildings
• Protocol and criteria? – Public dataset, – Avoiding risk to over-‐fipng/op<mizing to the data
![Page 43: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/43.jpg)
Detector evalua<ons
matchesallmatchescorrectprecision
##
=
A
BB
homography
Two points are correctly matched if T=40%
TBABA>
∪
∩
encescorrespondtruthgroundmatchescorrectrecall
##
=
![Page 44: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/44.jpg)
Previous Evalua<ons • 2D Scene – Homography
– C. Schmid, R. Mohr, and C. Bauckhage, “Evalua<on of interest point detectors,” IJCV, 2000.
– K. Mikolajczyk and C. Schmid, “A performance evalua<on of local descriptors,” CVPR, 2003.
– T. Kadir, M. Brady, and A. Zisserman, “An affine invariant method for selec<ng salient regions in images,” in ECCV, 2004.
– K. Mikolajczyk, T. Tuytelaars, C. Schmid, A. Zisserman, J. Matas, F. Schaffalitzky,T. Kadir, and L. Van Gool, “A comparison of affine region detectors,” IJCV, 2005.
– A. Haja, S. Abraham, and B. Jahne, Localiza<on accuracy of region detectors, CVPR 2008
– T. Dickscheid, FSchindler, Falko, W. Förstner, Coding Images with Local Features, IJCV 2011
![Page 45: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/45.jpg)
Previous Evalua<ons • 3D Scene -‐ epipolar constraints
– F. Fraundorfer and H. Bischof, “Evalua<on of local detectors on non-‐planar, scenes,” in AAPR, 2004.
– P. Moreels and P. Perona, “Evalua<on of features detectors and descriptors based on 3D objects,” IJCV, 2007.
– S. Winder and M. Brown, “Learning local image descriptors,” CVPR, 2007,2009.
– Dahl, A.L., Aanæs, H. and Pedersen, K.S. (2011): Finding the Best Feature Detector-‐Descriptor Combina<on. 3DIMPVT, 2011.
![Page 46: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/46.jpg)
Recent Evalua<ons
• Recent detectors O. Miksik and K. Mikolajczyk, Evalua<on of Local Detectors and Descriptors for Fast Feature Matching, ICPR 2012
ECCV 2012 Modern features: … Detectors. 46/60
![Page 47: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/47.jpg)
Recent detector evalua<ons • Completeness (coverage), complementarity between
detectors
T. Dickscheid, FSchindler, Falko, W. Förstner, Coding Images with Local Features, IJCV 2011
– Completeness, Edgelap (Mikolajczyk), Salient (Kadir), MSER (Matas)
– Complementrity, MSER + SFOP
![Page 48: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/48.jpg)
Descriptor Evalua<ons
• Matching precision and recall
O. Miksik and K. Mikolajczyk, Evalua<on of Local Detectors and Descriptors for Fast Feature Matching, ICPR 2012
![Page 49: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/49.jpg)
Recent Descriptor Evalua<ons • Computa%on %mes for the different descriptors for 1000
SURF keypoints O. Miksik and K. Mikolajczyk, Evalua<on of Local Detectors and Descriptors for Fast Feature Matching, ICPR 2012 J. Heinly E. Dunn, J-‐M. Frahm, Compara<ve Evalua<on of Binary Features, ECCV2012
![Page 50: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/50.jpg)
Previous Evalua<ons • Image/object categories
– K. Mikolajczyk, B. Leibe, and B. Schiele, “Local features for object class recogni<on,” in ICCV, 2005
– E. Seemann, B. Leibe, K. Mikolajczyk, and B. Schiele, “An evalua<on of local shape-‐based features for pedestrian detec<on,” in BMVC, 2005.
– M. Stark and B. Schiele, “How good are local features for classes of geometric objects,” in ICCV, 2007.
– K. E. A. van de Sande, T. Gevers and C. G. M. Snoek, Evalua<on of Color Descriptors for Object and Scene Recogni<on. CVPR, 2008.
![Page 51: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/51.jpg)
Approach • Bags-‐of-‐features
1. Interest point / region detector 2. Descriptors 3. K-‐means clustering (4000 clusters) 4. Histogram of cluster occurrences (NN assignment) 5. Chi-‐square distance and RBF kernel for KDA or SVM classifier
• J. Zhang and M. Marszalek and S. Lazebnik and C. Schmid, Local Features and Kernels for Classifica<on of Texture and Object Categories: A Comprehensive Study, IJCV, 2007 • K. E. A. van de Sande, T. Gevers and C. G. M. Snoek, Evalua<on of Color Descriptors for Object and Scene Recogni<on. CVPR, 2008
![Page 52: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/52.jpg)
Evalua<on • PASCAL VOC measures
– Average precision for every object category – Mean average precision AP Category
precision
recall
AP output=>
![Page 53: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/53.jpg)
MAP #dimensions density
MAP Ranking color/gray, density, dimensionality ...
• SIFT s<ll dominates (Histograms of gradient loca<ons and orienta<ons) • Opponent chroma<c space (normalized red-‐green, blue-‐yellow, and intensity Y
![Page 54: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/54.jpg)
Grayvalue descriptors
• Observa<ons
• Color improves • All based on histograms of gradient loca<ons and orienta<ons • Dimensionality not much correlated with the performance • Density Strongly correlated (the more the beeer) • Results biased by density • Implementa<on details maeer
MAP Ranking density #dimensions
![Page 55: Modern features-part-2-descriptors](https://reader034.vdocuments.us/reader034/viewer/2022042607/555a7669d8b42a972b8b531a/html5/thumbnails/55.jpg)
Break !