Presentation is loading. Please wait.

Presentation is loading. Please wait.

Taxonomic classification for web- based videos Author: Yang Song et al. (Google) Presenters: Phuc Bui & Rahul Dhamecha.

Similar presentations


Presentation on theme: "Taxonomic classification for web- based videos Author: Yang Song et al. (Google) Presenters: Phuc Bui & Rahul Dhamecha."— Presentation transcript:

1 Taxonomic classification for web- based videos Author: Yang Song et al. (Google) Presenters: Phuc Bui & Rahul Dhamecha

2 1. Introduction

3 Taxonomic classification for web-based videos

4 Web-based Video Classification Web-based Video (e.g. Youtube) – Over 800 million unique users visit / month – Over 4 billion hours of video are watched / month – 72 hours of video are uploaded / minute Classification – Improve User experience – Increase Website profit

5 What’s interesting? Large-scale classification – Taxonomy of categories – Unlimited domain Combined Approach – Text Labeled Web documents Labeled Video – Video Content-based features

6 Overview Approach Multi-labels Classification – One classifier for each category Classifiers – Text-based Classifier from Web-based Documents – Combined Classifier Text-based Classifier Video content-based features

7 2. Algorithms

8 TAXONOMIC CLASSIFICATION: - THE VARIOUS CATEGORIES.

9 TRAINING SET OF EACH CATEGORY

10 Pre-trained text based classifiers of each category used for porting videos Labeled Video data is used for training these classifiers No. of Classifiers = No. of Categories Ada-boosting is deployed to aggregate these weak classifiers to a Strong Classifier MIGRATION FROM TEXT TO VIDEO

11 Feature Extraction Text Based Features. President Obama: the Real Mitt Romney - Denver, Colorado Title Description Keywords

12 Content Based Features Moments from multi-scale analysis Color Histogram Mean, variance of each channel. Difference between mean of center and boundary

13 Content Based Features contd… Edge Detection Canny Edge Detection Algorithm

14 Content Based Features contd… Color Motion Features Cosine Difference of the histograms of subsequent frames.

15 Content Based Features contd… Shot Boundary Features Types Hard Cut Fade Dissolve Wipe

16 Hard Cut instantaneous transition from one scene to the next

17 Fade A Fade which is a gradual between a scene and a constant image (fade-out) or between a constant image and a scene (fade-in).

18 Dissolve A Dissolve is a gradual transition from one scene to another in which the first scene fade-out and the second scene fade-in. so it is a combination of fade-in and fade-out.

19 Wipe A Wipe is a gradual transition in which a line move across the screen, with the new scene appearing behind the line.

20 Integration Labled Videos Text Based Feature Extraction Apply Pre- trained Text Classifiers F Score from Classifiers Labled Videos Content Based Feature Extraction F Score and Content Based Features are combined A new Classifier is trained.

21 3. Experiments

22 Data 5789 videos 9087 labels 565 categories 80% training 20% evaluation

23 Evaluation

24 Results Sample videos

25 Results 80-category classifiers 1037-category classifiers

26 Results

27 Adaption + Content- based features classifiers Content-based features- only classifiers

28 4. Conclusion & Dicussion Video features – Content-based – Associated texts Web-documents based text classifier Semi-supervised learning Image-based classifiers – ImageNet


Download ppt "Taxonomic classification for web- based videos Author: Yang Song et al. (Google) Presenters: Phuc Bui & Rahul Dhamecha."

Similar presentations


Ads by Google