Private:copyDetection: Notes - Revision history

Mhefeeda at 20:30, 29 December 2010

2010-12-29T20:30:44Z

2010-12-29T20:30:17Z

2010-12-29T20:29:44Z

2010-12-29T20:29:18Z

2010-12-28T21:20:27Z

2010-10-01T22:30:41Z

New page: == Video Copy Detection using Optical Flow == * any previous work? == 3D Video Copy Detection == * ideas, previous works?

New page

== Video Copy Detection using Optical Flow ==

* any previous work?

== 3D Video Copy Detection ==

* ideas, previous works?

@@ Line 7: / Line 7: @@
 * Extract local features (e.g., SIFT) from I frames
-* Cluster these features into K clusters using for example K-means method (need to try for different values of K)
+* Cluster these features into K clusters using for example K-means method (need to try different values of K)
 * To create signatures

@@ Line 10: / Line 10: @@
 * To create signatures
 ** Divide a video into groups (may be GOP)
 **Extract local features from I frames
@@ Line 16: / Line 15: @@
 **Normalize the probabilities so that they sum to 1
 **Use these (k) probabilities as a signature.
 **In addition, we extract motion vectors from non I-frames in the GOP.
 **Quantize these motion vectors into fixed number of bins, say B
 **Build a histogram on these bins
 **Normalize and compute probabilities (vector of size B).
 **Now, use a combined signature from local features (K vector) and motion info (B vector).
 *For comparing :
@@ Line 32: / Line 28: @@
 *Notes:
 **Signature creation can be done on a moving window, i.e., shifting with each frame (computationally expensive though).
 ** Later, we can create another level of abstraction to improve performance: Use the K vector (local features) and build a topic model on top of it using for example LDA. That is, each k-vector will be used as a word. The topic model will identify the collection of words that commonly occur together (which is called a topic).

@@ Line 16: / Line 16: @@
 **Normalize the probabilities so that they sum to 1
 **Use these (k) probabilities as a signature.
+--
 **In addition, we extract motion vectors from non I-frames in the GOP.
 **Quantize these motion vectors into fixed number of bins, say B

@@ Line 1: / Line 1: @@
 == Video Copy Detection ==
-'''Early Idea'''
+'''Early Idea (discussed with Dr. Wael Abd-Almageed)'''
 * Collect large set of videos (may be from TREC)
@@ Line 12: / Line 12: @@
 ** Divide a video into groups (may be GOP)
-** Extract local features from I frames
+**Extract local features from I frames
-** Map these features to the K clusters (prob value for each cluster)
+**Map these features to the K clusters (prob value for each cluster)

← Older revision		Revision as of 21:20, 28 December 2010
Line 1:		Line 1:
		+	== Video Copy Detection ==

−	~~== Video Copy Detection using Optical Flow ==~~	+	'''Early Idea'''

−	* ~~any previous work?~~	+	* Collect large set of videos (may be from TREC)
		+
		+	* Extract local features (e.g., SIFT) from I frames
		+
		+	* Cluster these features into K clusters using for example K-means method (need to try for different values of K)
		+
		+	* To create signatures
		+
		+	** Divide a video into groups (may be GOP)
		+	** Extract local features from I frames
		+	** Map these features to the K clusters (prob value for each cluster)
		+
		+	* Extract motion vectors from