crowdsourced automatic zoom and scroll for video
play

Crowdsourced Automatic Zoom and Scroll for Video Retargeting Axel - PowerPoint PPT Presentation

Crowdsourced Automatic Zoom and Scroll for Video Retargeting Axel Carlier, Wei Tsang Ooi NUS (Singapore) Vincent Charvillat, Romulus Grigoras, Geraldine Morin IRIT (Toulouse, France) 1 iPhone 4 Video retargeting: Making a large video fit


  1. Crowdsourced Automatic Zoom and Scroll for Video Retargeting Axel Carlier, Wei Tsang Ooi NUS (Singapore) Vincent Charvillat, Romulus Grigoras, Geraldine Morin IRIT (Toulouse, France) 1

  2. iPhone 4 Video retargeting: Making a large video fit into a smaller screen and available with network capacities. 2

  3. One simple way: Scale down the video to the resolution of the screen 3

  4. Example 4

  5. Drawback: Important details may not be visible 5

  6. What makes a good video retargeting ? - Good comprehension of the video content - The video is aesthetically satisfying 6

  7. Important region Zoom 7

  8. Idea : Zoom and scroll to show only interesting regions of the video 8 8

  9. Problem: how to find the region automatically ? 9

  10. Approaches using Content Analysis Saliency map Motion detection Too many regions Liu, Gleicher MM 06 Avidan, Shamir Commun. ACM 09 10 10

  11. Speech Recognition + Natural Language processing Object recognition

  12. Our idea: Crowdsourcing Identifying regions of interest by gathering implicit input from users. 12 12

  13. Viewer Retargeted traces video Passive Active Viewers Viewers

  14. Use zoomable video 14 14

  15. Interface 15 15

  16. Example of user interaction 16 16

  17. Crowdsourcing ● Tutorial: how to use the interface ? ● Magic videos : – HD Videos : 1920 × 1080 pixels – Fixed camera – Obvious ROIs : magician's hands, cards, dice... ● Between 7 and 12 viewers for each video ● 11,183 interaction events logged 17 17

  18. Automatic Generation of Retargeted Video Heatmap ROIs Shot 2 Shots Shot 1 Final Video 18 18

  19. Building Heatmaps 19 19

  20. Building Heatmaps 20 20

  21. Building Heatmaps 21 21

  22. Analyzing Heatmaps Here draw gaussians in 3d with matlab GMM (Gaussian Mixture model) -K -w i Mean Shift -m i - ∑ i 22 22

  23. Finding Modes Mean-Shift: Clustering algorithm (Comaniciu, ICCV 02) 23 23

  24. Determining ROI size Minimum Covariance Determinant (MCD) 24 24

  25. Building a ROI Dynamics Graph

  26. Cutting the graph into shots

  27. Shots selection Shot 1 Shot 3 Shot 2

  28. Result video 28 28

  29. Integrating Reframing techniques ● Bottom-up reframing ● Type of shot: fixed, zooming or dolly ● Shot level: stabilization according to its type ● Inter-shot level: transitions and reestablishing shots 29 29

  30. Dolly shot Shot 3 30 30

  31. Shots 31 31

  32. Shot stabilization 32 32

  33. Transitions 33 33

  34. Reestablishing shots 34 34

  35. Final Result 35 35

  36. Results validation ● 3 poor videos: ● User interaction ( user ) ● Retargeted version without reframing techniques ( noRT ) ● Original version scaled down ( nozoom ) ● Retargeted version with reframing techniques ( crowdsourced ) ● Ground truth ( expert ) 36 36

  37. Ground Truth 37 37

  38. Protocol ● 48 participants divided into 3 categories ● User – crowdsourced – expert (18) ● NoRT – crowdsourced – expert (18) ● Nozoom – crowdsourced – expert (12) ● 3 questions were asked to the participants 38 38

  39. Rate the video editing of the video 5 4 3 2 1 0 39 39 NoRT = retargeted version without reframing techniques

  40. Is the video editing reasonable ? 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 40 40

  41. Does the video manage to convey important information ? 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 41 41

  42. Summary -Gather implicit input from users -No content analysis - In our examples : less than 12 viewers are enough to detect ROIs

  43. Future work ● Explore alternative methods for intermediary steps: ● Modelling heatmaps not as a GMM ● Adding cinematographic rules ● Classify users into different profiles and generate a retargeted video for each profile 43 43

  44. Questions ? 44 44

  45. Results 45 45

  46. Using aesthetics Liu, Chen, Wolf and Cohen-Or. Optimizing Photo Composition, Computer Graphic Forum Luo, Yi wen and Tang, Xiaoou. Photo and Video Quality Evaluation : Focusing on the Subject, ECCV 08 46

  47. MCD Covariance 47

  48. MCD Covariance 48

  49. Gym Video Retargeted 49

  50. Crowdsourcing Shamma, Shaw, Shafton, Liu. Watch what I watch, MIR 07 50

  51. Overview ● Video retargeting ● Zoomable video ● Finding users' interests ● Creating shots ● Integrating reframing techniques ● Results validation 54 54

  52. 55 55 http://en.wikipedia.org/wiki/File:UHDV.svg

  53. 960 x 640 iPhone 4 56 56 http://en.wikipedia.org/wiki/File:UHDV.svg

  54. 57 57

  55. 58 58

  56. Approaches using Content Analysis Saliency map Motion detection Liu, Gleicher MM 06 Avidan, Shamir Commun. ACM 09 59 59

  57. Crowdsourcing Heatmap Hotspots Shot 2 Shots Shot 1 Final Video 60 60

  58. Creating Heatmaps ● Modelization of ROIs as a GMM (Gaussian Mixture Model) 61 61

  59. Creating Heatmaps ● Modelization of ROIs as a GMM (Gaussian Mixture Model) 62 62

  60. Building a ROI dynamics graph 63 63

  61. Minimal spanning tree 64 64

  62. Cutting the tree into shots 65 65

  63. Shots selection Shots are selected according to their popularity: 66 66

Recommend


More recommend