際際滷

際際滷Share a Scribd company logo
Unmasking the abnormal events in video
朱語所鍵 覈 2018-1
 覦 觜  語 郁規 危覩
覓伎 朱語瑚
 Radu Tudor Ionescu, Sorina Smeureanu, Bogdan Alexe, Marius Popescu; The IEEE International
Conference on Computer Vision (ICCV), 2017, pp. 2895-2903
 Video 螳 Scene Abnormal讌 覿襯  螻襴讀
 Text  Author Verification 一 Unmasking(Koppel et al.) 蠍磯 Video Frame 
 Unmasking   data 螳 襯 豸′蠍 , 螳 discriminant Feature襯 蟇壱 螳覃
 Classification Accuracy螳 螳 襯 蟯谿
Unmasking
Unmasking
Unmasking
Unmasking
Overall Framework
2*w螳 Frame 觸.
 w螳 normal, れ w螳 abnormal襦 Labeling.
Motion Feature Appearance Feature襯 豢豢.
豢豢 Feature襯 Linear Model襦 Classify.
Classifier Train Accuracy襯 豸′.
螳 Discriminant Feature 豢   れ Linear
Classifier襦 覲企碁.  Accuracy螳 螳
蠏 Accuracy襯 Plot.
Overall Framework
2*w螳 Frame 觸.
 w螳 normal, れ w螳 abnormal襦 Labeling.
Motion Feature Appearance Feature襯 豢豢.
豢豢 Feature襯 Linear Model襦 Classify.
Classifier Train Accuracy襯 豸′.
螳 Discriminant Feature 豢   れ Linear
Classifier襦 覲企碁.  Accuracy螳 螳
蠏 Accuracy襯 Plot.
Feature???
Feature Extraction
Classification
Feature???
Feature Extraction
Classification
Motion Feature
x
y
t
Motion Feature
x
y
t
10
10
5
Spatio  Temporal
3D - Gradient
Appearance Feature
Appearance Feature
Linear Classification
Normal
Abnormal
Plotting
Thank You!!

More Related Content

[Paper Review] Unmasking the abnormal events in video