1 / 25

Groups of Adjacent Contour Segments for Object Detection

Vittorio Ferrari Loic Fevrier Frederic Jurie Cordelia Schmid. Groups of Adjacent Contour Segments for Object Detection. ?. Problem: object class detection & localization. Training. Focus : classes with characteristic shape. Testing. Features: pairs of adjacent segments (PAS).

vian
Download Presentation

Groups of Adjacent Contour Segments for Object Detection

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Vittorio Ferrari Loic Fevrier Frederic Jurie Cordelia Schmid Groups of Adjacent Contour Segments for Object Detection

  2. ? Problem: object class detection & localization Training Focus: classes with characteristic shape Testing

  3. Features: pairs of adjacent segments (PAS) Contour segment network [Ferrari et al. ECCV 2006] edgels extracted with Berkeley boundary detector 2) edgel-chains partitioned into straight contour segments 3) segments connected at edgel-chains’ endpoints and junctions

  4. PAS descriptor: Features: pairs of adjacent segments (PAS) segments connected in the network PAS = groups of two connected segments • encodes geometric properties of the PAS • scale and translation invariant • compact, 5D

  5. Example PAS Why PAS ? + can cover pure portions of the object boundary Features: pairs of adjacent segments (PAS) + intermediate complexity: good repeatability-informativeness trade-off + scale-translation invariant + connected: natural grouping criterion (need not choose a grouping neighborhood or scale)

  6. PAS codebook Based on descriptors, cluster PAS into types a few of the most frequent types based on 10 outdoor images (5 horses and 5 background). types based on 15 indoor images (bottles) • Frequently occurring PAS have intuitive, natural shapes • As we add images, number of PAS types converges to just ~100 • Very similar codebooks come out, regardless of source images + general, simple features. We use a single, universal codebook (1st row) for all classes

  7. Window descriptor 1. Subdivide window into tiles. 2. Compute a separate bag of PAS per tile 3. Concatenate these semi-local bags [Lazebnik et al. CVPR 2006]; [Dalal and Triggs CVPR 2005] + distinctive: records which PAS appear where weight PAS by average edge strength + flexible: soft-assign PAS to types rather coarse tiling + fast to compute using Integral Histograms

  8. 4. Collect negative example descriptors: slide window over negative training images Training 1. Learn mean positive window dimensions 2. Determine number of tiles T 3. Collect positive example descriptors

  9. Training 5. Train a linear SVM Here a few of the top weighted descriptor vector dimensions (= 'PAS + tile'): + lie on object boundary (= local shape structure common to many training examples)

  10. Testing 1. Slide window of aspect ratio , at multiple scales 2. SVM classify each window + non-maxima suppression detections

  11. (missed and FP) Results – INRIA horses Dataset: ~ Jurie and Schmid, CVPR 2004 170 positive + 170 negative images (training = 50 pos + 50 neg) wide range of scales; clutter + tiling brings a substantial improvement optimum at T=30 -> keep this setting on all other experiments + works well: 86% det-rate at 0.3 FPPI (with 50 pos + 50 neg training images)

  12. (missed and FP) Results – INRIA horses Dataset: ~ Jurie and Schmid, CVPR 2004 170 positive + 170 negative images (training = 50 pos + 50 neg) wide range of scales; clutter + PAS better than any IP all interest point (IP) comparisons with T=10, and 120 feature types, (= optimum over INRIA horses, and ETHZ Shape Classes; all IP codebooks are class-specific)

  13. Results – Weizmann-Shotton horses Dataset: Shotton et al., ICCV 2005 327 positive + 327 negative images (training = 50 pos + 50 neg) no scale changes; modest clutter Shotton’s EER - exact comparison to Shotton et al.: use their images and search at a single scale - PAS same performance (~92% precision-recall EER), but: + no need for segmented training images (only bounding-boxes) + can detect objects at multiple scales (see other experiments)

  14. Results – ETHZ Shape Classes Dataset: Ferrari et al., ECCV 2006 255 images, over 5 classes training = half of positive images for a class + same number from the other classes (1/4 from each) testing = all other images large scale changes; extensive clutter

  15. Missed Results – ETHZ Shape Classes Dataset: Ferrari et al., ECCV 2006 255 images, over 5 classes training = half of positive images for a class + same number from the other classes (1/4 from each) testing = all other images large scale changes; extensive clutter

  16. Apple logos Bottles Giraffes Mugs Swans Results – ETHZ Shape Classes + mean det-rate at 0.4 FPPI = 79% + class specific IP codebooks + PAS >> I.P for apple logos, bottles, mugs PAS ~= IP for giraffes (texture!) PAS < IP for swan + overall best IP: Harris-Laplace

  17. Results – Caltech 101 Results – Caltech 101 Dataset: Fei-Fei et al., GMBV 2004 42 anchor, 62 chair, 67 cup images train = half + same number of caltech101 background testing = other half pos + same number of background scale changes; only little clutter

  18. Results – Caltech 101 Dataset: Fei-Fei et al., GMBV 2004 On caltech101’s anchor, chair, cup: + PAS better than Harris-Laplace + mean PAS det-rate at 0.4 FPPI: 85%

  19. Apple logos Bottles Giraffes Mugs Swans Comparison to Dalal and Triggs CVPR 2005

  20. INRIA horses Shotton horses Caltech anchors Caltech chairs Caltech cups Comparison to Dalal and Triggs CVPR 2005 + overall mean det-rate at 0.4 FPPI: PAS 82% >> HoG 58% PAS >> HoG for 6 datasets PAS ~= HoG for 2 datasets PAS < HoG for 2 datasets

  21. Generalizing PAS to kAS kAS: any path of length k through the contour segment network segments connected in the network 4AS 3AS • scale+translation invariant descriptor with dimensionality 4k-2 • k = feature complexity; higher k -> more informative, but less repeatable kAS • overall mean det-rates (%) 1AS PAS 3AS 4AS 0.3 FPPI 69 77 64 57 0.4 FPPI 76 82 70 64 PAS do best !

  22. Conclusions Connected local shape features for object class detection Experiments on 10 diverse classes from 4 datasets show: + bettersuited than interest points for these shape-based classes + PAS have the best intermediate complexity among kAS + object detectordeals with clutter, scale changes, intra-class variability + object detector compares favorably to HoG-based one - fixed aspect-ratio window: sometimes inaccurate bounding-boxes - single viewpoint

  23. Model • collection of PAS and their spatial variability • only common boundary Current work: detecting object outlines Training: learn the common boundaries from examples

  24. 3. vote for translation + scale initializations 4. match deformable thin-plate spline based on deterministic annealing Outline object in test image, without segmented training images ! Current work: detecting object outlines Detection on a new image 1. detect edges 2. match PAS based on descriptors

  25. A few preliminary results

More Related