synchronizing multi-user photo galleries with mrf

31
Synchronizing Multi- User Photo Galleries with MRF Emanuele Sansone, Giulia Boato Minh-Son Dao DISI, University of Trento UIT-HCM Trento, Italy HCMC, Viet Nam [email protected] , [email protected] [email protected]

Upload: multimediaeval

Post on 05-Aug-2015

38 views

Category:

Software


2 download

TRANSCRIPT

Page 1: Synchronizing Multi-User Photo Galleries with MRF

Synchronizing Multi-User Photo Galleries with MRF

Emanuele Sansone, Giulia Boato Minh-Son Dao DISI, University of Trento UIT-HCM Trento, Italy HCMC,

Viet Nam [email protected], [email protected] [email protected]

Page 2: Synchronizing Multi-User Photo Galleries with MRF

Synchronization and Clustering

Multiple Galleries

Feature Extraction

Offset Estimation

Synchronized Galleries

Page 3: Synchronizing Multi-User Photo Galleries with MRF

Stereo Matching I

Stereo camera

Acquired Images

Page 4: Synchronizing Multi-User Photo Galleries with MRF

Stereo Matching II

Page 5: Synchronizing Multi-User Photo Galleries with MRF

Observed node = pixel intensity

Latent node = possible disparity for a given node

Potentials associated to each link in the network

Inference algorithm

Disparity

Stereo Matching: Model

Page 6: Synchronizing Multi-User Photo Galleries with MRF

• Pixels Photos• Pixel Intensitities Image features (Timestamp, GPS, SURF, color

histograms)• States: possible diparities States: possible offsets

Stereo Matching vs. Photo Galleries Synchronization

Page 7: Synchronizing Multi-User Photo Galleries with MRF

Synchronization: States

Page 8: Synchronizing Multi-User Photo Galleries with MRF

Synchronization: Potentials

Page 9: Synchronizing Multi-User Photo Galleries with MRF

Synchronization: Potentials

Page 10: Synchronizing Multi-User Photo Galleries with MRF

Synchronization: Potentials

Page 11: Synchronizing Multi-User Photo Galleries with MRF

Synchronization: Potentials

Page 12: Synchronizing Multi-User Photo Galleries with MRF

Synchronization: Potentials

Page 13: Synchronizing Multi-User Photo Galleries with MRF

Synchronization: Potentials

Page 14: Synchronizing Multi-User Photo Galleries with MRF

Synchronization: Final Model

Page 15: Synchronizing Multi-User Photo Galleries with MRF

Synchronization: Inference

Page 16: Synchronizing Multi-User Photo Galleries with MRF

Clustering

• K-means algorithm• 3 runs

Page 17: Synchronizing Multi-User Photo Galleries with MRF

Clustering

• K-means algorithm• 3 runs

Color Structure Descriptors(CSD - 64 values) + Local Binary Patterns 3x3 (LBP - 18 values each block)

Page 18: Synchronizing Multi-User Photo Galleries with MRF

Clustering

• K-means algorithm• 3 runs

Color Structure Descriptors(CSD - 64 values) + Local Binary Patterns 3x3 (LBP - 18 values each block)

PCA Color Structure Descriptors (6 values)

Page 19: Synchronizing Multi-User Photo Galleries with MRF

Clustering

• K-means algorithm• 3 runs

Color Structure Descriptors(CSD - 64 values) + Local Binary Patterns 3x3 (LBP - 18 values each block)

PCA Color Structure Descriptors (6 values)

PCA Color Structure Descriptors (6 values) + TIME

Page 20: Synchronizing Multi-User Photo Galleries with MRF

Results: Synchronization

Dataset Precision Accuracy

Vancouver 0.35 0.86

London 0.25 0.89

Page 21: Synchronizing Multi-User Photo Galleries with MRF

Results: Synchronization

Dataset Precision Accuracy

Vancouver 0.35 0.86

London 0.25 0.89

• Good performance in terms of accuracy

Page 22: Synchronizing Multi-User Photo Galleries with MRF

Results: Synchronization

Dataset Precision Accuracy

Vancouver 0.35 0.86

London 0.25 0.89

• Good performance in terms of accuracy• Only 1/4 galleries correctly synchronized

Page 23: Synchronizing Multi-User Photo Galleries with MRF

Results: ClusteringDataset Run Features Jaccard Index

Vancouver

1 CSD-64+LBP3x3 0.1673

2 CSD-6 0.1382

3 CSD-6+TIME 0.1315

London

1 CSD-64+LBP3x3 0.1287

2 CSD-6 0.0742

3 CSD-6+TIME 0.0885

Page 24: Synchronizing Multi-User Photo Galleries with MRF

Results: ClusteringDataset Run Features Jaccard Index

Vancouver

1 CSD-64+LBP3x3 0.1673

2 CSD-6 0.1382

3 CSD-6+TIME 0.1315

London

1 CSD-64+LBP3x3 0.1287

2 CSD-6 0.0742

3 CSD-6+TIME 0.0885

• Performance are decreased by using only color descriptors

Page 25: Synchronizing Multi-User Photo Galleries with MRF

Results: ClusteringDataset Run Features Jaccard Index

Vancouver

1 CSD-64+LBP3x3 0.1673

2 CSD-6 0.1382

3 CSD-6+TIME 0.1315

London

1 CSD-64+LBP3x3 0.1287

2 CSD-6 0.0742

3 CSD-6+TIME 0.0885

• Performance are decreased by using only color descriptors• In the Vancouver dataset time doesn’t help in clustering

Page 26: Synchronizing Multi-User Photo Galleries with MRF

Results: ClusteringDataset Run Features Jaccard Index

Vancouver

1 CSD-64+LBP3x3 0.1673

2 CSD-6 0.1382

3 CSD-6+TIME 0.1315

London

1 CSD-64+LBP3x3 0.1287

2 CSD-6 0.0742

3 CSD-6+TIME 0.0885

• Performance are decreased by using only color descriptors• In the Vancouver dataset time doesn’t help in clustering• In the London dataset time becomes a more reliable feature

Page 27: Synchronizing Multi-User Photo Galleries with MRF

Lessons Learned

Synchronization:Only 1/4 galleries correctly synchronized

Page 28: Synchronizing Multi-User Photo Galleries with MRF

Lessons Learned

Synchronization:Only 1/4 galleries correctly synchronized• new features (e.g. text tags, social network data)

Page 29: Synchronizing Multi-User Photo Galleries with MRF

Lessons Learned

Synchronization:Only 1/4 galleries correctly synchronized• new features (e.g. text tags, social network data)• “no association” between a pair of images

Page 30: Synchronizing Multi-User Photo Galleries with MRF

Lessons Learned

Synchronization:Only 1/4 galleries correctly synchronized• new features (e.g. text tags, social network data)• “no association” between a pair of images

Clustering:Time is important!

Page 31: Synchronizing Multi-User Photo Galleries with MRF

Thank you