行为识别

学习文章：going deeper into action recognition : a survey。samitha herath 等。

传统方法：

                 wholistic:

                 local :    sparse   stip cuboid hog3d

                              dense    trajectory

                               描述方法：（1）bow (zeroth order)  (2)FV (zero,one,two,orderth information) (3)vlad(zero,one                                                               orderth information)

THUMOS challenge 2014 总结：

（1）dense trajectory features : 10 teams used DTF (9 improved DTF, 1 basic DTF)

（2）convolutional nerual netwroks: 4 teams

（3）beyond low-level features:  body and facial features; audio; saliency features; shot boundary detection

（4）classification：10 teams used svm,1 extreme learning, no fully conneted cnn

（5）9 teams used fisher vector