Motion-Based Selection of Relevant Video Segments for Video Summarization
read more
Citations
Video Summarization: Techniques and Classification
System and method for video summarization
Stained-glass visualization for highly condensed video summaries
Video summarisation for surveillance and news domain
Hysteroscopy video summarization and browsing by estimating the physician's attention on video segments.
References
Stochastic Relaxation, Gibbs Distributions, and the Bayesian Restoration of Images
Detecting changes in signals and systems—a survey
Robust multiresolution estimation of parametric motion models
Comparison of video shot boundary detection techniques
A user attention model for video summarization
Related Papers (5)
Frequently Asked Questions (12)
Q2. What is the reason for the oversegmentation in the first third of the video?
The automatic segmentation leads to an oversegmentation in the first third of the video, probably due to the successive slight increases and decreases in the velocity of the camera following the actors in the corridor scenes.
Q3. What is the objective of the method in shots overview?
For an objective very close to video summarization, shots overview, the method in [5] relies on the nonlinear temporal modelling of waveletbased motion features.
Q4. What is the method used to compute the residual normal flow?
a continuous local motion measure is computed as a weighted mean, over a small spatial window, of the residual normal flow magnitude in order to obtain a more reliable motion information.
Q5. What is the goal of the proposed video segmentation method?
The goal of the proposed video segmentation method is only to detect changes in camera motion and not to identify the nature of this motion.
Q6. What is the way to characterize the dynamic content of the video?
Once temporal units of the processed video are identified, one way to characterize their dynamic content would be to consider again parametric motion models (e.g. 2D affine or quadratic motion models).
Q7. What is the study for the detection of a sequence of home activities in a video?
The study described in [18] for the detection of a sequence of home activities in a video relies on segmenting moving objects and detecting temporal discontinuities in the successive optical flow fields.
Q8. What is the basic unit of the video summarization method?
The video elementary unit considered in this method is simply the frame, which can be restrictive when trying to detect temporal semantic events.
Q9. What is the main requirement to carry on with the second step?
In all their experiments, the ability of the method to provide homogeneous segments in terms of camera motion has been proved, which is the main requirement to carry on with the second step.
Q10. How accurate is the evaluation of the residual normal flow?
the accuracy of the evaluation of the residual normal flow is highly dependent on the norm of the spatial intensity gradient, and this accuracy increases with ||∇ The author(p, k)||.
Q11. How many segments are correctly classified as play?
With Table 1(b), the authors can see that among the segments classified as play by their algorithm, only 3% are no play segments (false alarms).
Q12. what is the temporal co-occurrences distribution of the sequence y?
The temporal co-occurrences distribution (y) of the sequence y is a matrix { (ν, ν ′|y)}(ν,ν ′)∈ 2 defined by:(ν, ν ′ | y) = K∑k=2 ∑ p∈R δ(ν, y(p, k)) · δ(ν ′, y(p, k − 1)), (3)where δ(i, j) is the Kronecker symbol (equal to 1 if i = j and to zero otherwise).