Presentation is loading. Please wait.

Presentation is loading. Please wait.

On-the-fly Specific Person Retrieval University of Oxford 24 th May 2012 Omkar M. Parkhi, Andrea Vedaldi and Andrew Zisserman.

Similar presentations


Presentation on theme: "On-the-fly Specific Person Retrieval University of Oxford 24 th May 2012 Omkar M. Parkhi, Andrea Vedaldi and Andrew Zisserman."— Presentation transcript:

1 On-the-fly Specific Person Retrieval University of Oxford 24 th May 2012 Omkar M. Parkhi, Andrea Vedaldi and Andrew Zisserman

2 Overview People “Barack Obama” “George Bush” “Courtney Cox” Ranked ShotsTextual Queries Search for: On-the-fly i.e. with no previous knowledge or model for these queries Large collection of un-annotated videos Large collection of un-annotated videos

3 Scrubs Data Set 12 Episodes from Seasons 1-5 and 8 5 hours of video data About 400k frames, partitioned into 5k shots About 300k near frontal face detections 768 x 576 MPEG2 format

4 Demo Search for “Courteney Cox” in Scrubs dataset. Steps: 1.Download example images from Google 1.Train a ranking function 1.Apply ranking function to video collection

5 DEMO

6 Demo- Scrubs Data Set

7

8 On the fly person retrieval system Text Query “Courteney Cox” Negative Training Images Video Collection Face Tracks Facial Features & Descriptors Facial Features & Descriptors Results ON-LINE PROCESSING OFF-LINE PROCESSING Google Image Search “Courteney Cox” Facial Features & Descriptors Fast Linear Classifier Ranking

9 Detection and Tracking Viola-Jones face detection on each frame Tracking measures “ connectedness ” of a pair of faces by point tracks intersecting both Doesn ’ t require contiguous detections No drift Faces clustered into tracks [Everingham et al. 2006, Apostoloff & Zisserman, 2007]

10 Scrubs Data Set 12 Episodes from Seasons 1-5 and 8 5 hours of video data About 400k frames, partitioned into 5k shots 300k face detections 6k face tracks 768 x 576 MPEG2 format

11 Detecting facial feature points Pictorial structure model Joint model of feature appearance and position [Felzenszwalb and Huttenlocher’2004, Everingham et al. 2006 ]

12 Face Appearance Representation  Affine transformation of face to canonical frame  Independent photometric normalization of parts  Represent gradients over circle centred on facial feature points  Feature descriptor is a 3849 dimensional vector [ Everingham et al. 2006 ]

13 Negative Training Images Combination of faces from Random downloaded images Labeled Faces in the Wild dataset Caltech Faces dataset About 16k face detections. Caltech 10, 000 Web Faces: http://www.vision.caltech.edu/Image_Datasets/Caltech_10K_WebFaces/ Labeled Faces in the Wild: http://vis-www.cs.umass.edu/lfw/

14 On-the-fly Person Retrieval Text Query “Courteney Cox” Negative Training Images Video Collection Face Tracks Facial Features & Descriptors Facial Features & Descriptors Results ON-LINE PROCESSING OFF-LINE PROCESSING Google Image Search “Courteney Cox” Facial Features & Descriptors Fast Linear Classifier Ranking

15 DEMO

16 Demo- Scrubs Data Set

17 TRECVid 2011 (IACC.1.B) About 200 hours of video data. 8k videos. MPEG4, 320x240 pixels 130k shots, About 3 million face detections 25,535 face tracks.

18 DEMO

19 Demo - TRECVid 2011 (IACC.1.B)

20 Facial attributes – FaceTracer project  Examples:  gender: male, female  age: baby, child, youth, middle age, senior  race: white, black, asian  smiling, mustache, eye-wear, hair colour N. Kumar, P. N. Belhumeur and S. K. Nayar, FaceTracer: A Search Engine for Large Collections of Images with Faces, European Conference on Computer Vision (ECCV), 2010 http://www.cs.columbia.edu/CAVE/projects/face_search/ Method person independent training set with attribute facial feature representation discriminative training of classifier for attribute

21 DEMO

22 Facial attributes – Glasses

23 DEMO Facial attributes – Beard

24 Facial attributes – Eyes Closed

25 Quantitative Performance - Scrubs Dataset Performance evaluation for 3 guest actors (Brendan Fraser, Courteney Cox and Michael J Fox) 12 dataset videos split into training and test sets (3 Training, 9 Testing) Annotations: Manual labeling of training and test set for each actor Manual labeling of positive training images from Google Negative training images from Caltech Faces dataset.

26 Quantitative Performance - Scrubs Dataset Retrieval Average Precision (AP) Training Examples SourceAverage Precision Positive Negative Brendan Fraser Courteney Cox Michael J Fox Scrubs 0.560.880.49 GoogleScrubs0.250.620.52 GoogleCaltech0.410.560.57

27 Quantitative Performance - Scrubs Dataset Using more training data per track Training Examples Source # samples per track Average Precision Positive Negative Brendan Fraser Courteney Cox Michael J Fox Scrubs Single0.560.880.49 Scrubs Multiple0.60.880.53

28 Future Work Exploring sources for positive examples Better feature representations Combination of attributes and identities

29  Any Questions?


Download ppt "On-the-fly Specific Person Retrieval University of Oxford 24 th May 2012 Omkar M. Parkhi, Andrea Vedaldi and Andrew Zisserman."

Similar presentations


Ads by Google