In network aggregation using efficient routing techniques for event driven sensor network
Saliency Aggregation: A Data-driven Approach
description
Transcript of Saliency Aggregation: A Data-driven Approach
![Page 1: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/1.jpg)
Saliency Aggregation: A Data-driven Approach
Long Mai Yuzhen Niu Feng LiuDepartment of Computer Science, Portland
State UniversityPortland, OR, 97207 USA
![Page 2: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/2.jpg)
Outline
• Introduction• Saliency Aggregation• Experiments• Conclusion
![Page 3: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/3.jpg)
Introduction
• Two major observations– Different methods perform differently in saliency
analysis.– The performance of a saliency analysis method
varies with individual images.
![Page 4: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/4.jpg)
Introduction
![Page 5: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/5.jpg)
Introduction
• Aggregation advantages– Considers the performance gaps among individual
saliency analysis methods and better determines their contribution in aggregation.
– Considers that the performance of each individual saliency analysis method varies over images and is able to customize an appropriate aggregation model to each input image.
![Page 6: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/6.jpg)
Saliency Aggregation
• Standard Saliency Aggregation– Given a set of m saliency maps {Si || 1 ≤ i ≤ m}
computed from an image I.– The aggregated saliency value S(p) at pixel p of I
– Three different options for the function ζ
![Page 7: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/7.jpg)
Saliency Aggregation
![Page 8: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/8.jpg)
Saliency Aggregation
• Pixel-wise Aggregation– Associates each pixel p with a feature vector
– Using logistic model to model the posterior probability
![Page 9: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/9.jpg)
Saliency Aggregation
• Aggregation using Conditional Random Field– Model each pixel as a node.– The saliency label of each pixel depends not only
on its feature vector, but also the labels of neighboring pixels.
– The interactions within the pixels also depend on the features.
![Page 10: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/10.jpg)
Saliency Aggregation
• Aggregation using Conditional Random Field
![Page 11: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/11.jpg)
Saliency Aggregation
• Aggregation using Conditional Random Field
![Page 12: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/12.jpg)
Saliency Aggregation
• Image-Dependent Saliency Aggregation– Upgrade the aggregation model from P Y |X θ into
P Y |X θ I for each image I.– Find k nearest neighbors in the training set and
then trains a saliency aggregation model using these k images.
– Use the GIST descriptor to find similar images.
![Page 13: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/13.jpg)
Saliency Aggregation
![Page 14: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/14.jpg)
Experiments
• Dataset– FT image saliency dataset– Stereo Saliency dataset (SS)
![Page 15: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/15.jpg)
![Page 16: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/16.jpg)
Experiments
![Page 17: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/17.jpg)
Experiments
• Robustness of Saliency Aggregation
![Page 18: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/18.jpg)
Experiments
• Discussions– When all the individual methods fail to identify a
salient region in an image, saliency aggregation will usually fail too.
– The performance will sometimes be affected if the GIST method does not find similar images.
– The aggregation requires results from all the individual methods, it is slower than each individual one.
![Page 19: Saliency Aggregation: A Data-driven Approach](https://reader035.fdocuments.net/reader035/viewer/2022070420/56815e45550346895dccb9da/html5/thumbnails/19.jpg)
Conclusion
• We presented data-driven approaches to saliency aggregation that integrate saliency analysis results from multiple individual saliency analysis methods.
• Image-dependent CRF-based approach that considers the interaction among pixels, the performance gaps among individual saliency analysis methods, and the dependent of saliency analysis on individual image