An efficient algorithm for Co-segmentation
read more
Citations
iCoseg: Interactive co-segmentation with intelligent scribble guidance
Discriminative clustering for image co-segmentation
Asymptotics in Statistics–Some Basic Concepts
One-Shot Learning for Semantic Segmentation
Unsupervised Joint Object Discovery and Segmentation in Internet Images
References
Active shape models—their training and application
Fast approximate energy minimization via graph cuts
Fast approximate energy minimization via graph cuts
SIMPLIcity: semantics-sensitive integrated matching for picture libraries
SIMPLIcity: Semantics-sensitive Integrated Matching for Picture Libraries
Related Papers (5)
Frequently Asked Questions (14)
Q2. What is the objective function for the MRF?
Their objective function minimizes a combination of the penalties incurred by the MRF optimization in each image, and subtracts the similarity measure of the number of pairs of the same histogram buckets in the resulting two foreground features.
Q3. What is the rationale for a binaryspq?
Their rationale is that for each pixel p assigned as foreground in the first image, the authors offer a reward for also selecting (as part of the foreground in the second image) a pixel q which is similar to p. Similarity which is specified by a binaryspq depends on whether p and q belong to the same bucket, and may be allowed to vary in [0, 1] as a function of the likelihood of the match (p → q) detected by some feature extraction method.
Q4. What is the simplest interpretation of the cosegmentation model?
The simplest interpretation of the cosegmentation model, as discussed above, is that it encourages good and coherent segmentation of both images with an additional requirement of consistency between foreground histograms.
Q5. What is the way to optimize the image?
For |F1|, |F2| fixed, the optimization process seeks to maximize the number of pixel pairs (one from each image) with identical histogram buckets.
Q6. What is the procedure for finding the correlation coefficient of the two foreground regions?
This involves parameterizing λ, solving a single parametric max-flow procedure [22, 23], and finding the correlation coefficient of the two foreground regions for each breakpoint.
Q7. What is the optimal solution to (Co-Seg)?
Then the optimal solution to (Co-Seg) is achieved by setting xi = 1 for each pixel node in the source set S and every zi1k,i2k = 1 for each similarity node in the sink set T .Proof:
Q8. What is the property of the co-seg problem?
Due to Property 3.1, the authors can make use of a construction of an s, t graph G, where the solution to the s, t-cut problem will provide an optimal solution to the (Co-seg) problem.
Q9. Why is the cosegmentation algorithm so invariant to the other?
In addition, due to their choice of rewarding similarity in histogram features (see (5)), small changes in scale of the object (between images) do not have a significant impact on the empirical performance of the algorithm.
Q10. How many iterations did the algorithm take to solve?
This requires solving a sequence of graph cuts and the process terminates once the algorithm has converged or the number of iterations have been reached (number of iterations was set to 10).
Q11. What is the objective value of the (Co-Seg) problem?
which is precisely the objective value of the (Co-Seg) problem, when setting the x and z variables with corresponding nodes in S to 1.
Q12. What is the objective for the co-segmentation problem?
To achieve only the MRF segmentation for both images specified as (4), the authors can use a graph construction similar to the one described in [17], with either the 4-neighbor or the 8-neighbor or any other form of neighborhood topology used to describe the adjacency relationship between pixel-nodes.
Q13. What did the authors do to improve the cosegmentation model?
the authors in [4] extended many of these ideas further by incorporating local context, i.e., patterns characterizing the local color and edge configurations.
Q14. Why is the algorithm motivated by the carrot or stick philosophy?
Their approach is motivated from the carrot or stick philosophy – where rather than penalize the difference (distance) of the two foreground histograms, the authors reward their similarity (affinity).