PerfXLab : https://github.com/PerfXLab
Angela Yao https://angelayao.com/publications/
Vlad I. Morariu http://www.umiacs.umd.edu/~morariu/
https://www.zhihu.com/question/33272629
https://www.zybuluo.com/vivounicorn/note/872759
http://slideplayer.com/slide/9169560/
Learning Discriminative Representation for Skeletal Action Recognition Using LSTM Networks
http://www.uow.edu.au/~leiw/ http://www.uow.edu.au/~leiw/share_files/
https://github.com/jysung/activity_detection http://pr.cs.cornell.edu/humanactivities/
http://ravitejav.weebly.com/rolling.html
http://theorangeduck.com/page/neural-network-not-working
action recognition dataset: http://homepages.inf.ed.ac.uk/rbf/CVonline/Imagedbase.htm
http://www.icst.pku.edu.cn/struct/icmew2017/result.html
Learning features combination for human action recognition from skeleton sequences https://github.com/dluvizon/harskel Interpretable 3D Human Action Analysis with Temporal Convolutional Networks https://github.com/TaeSoo-Kim/TCNActionRecognition
http://www.di.ens.fr/willow/research/ltc/
https://hongsong-wang.github.io/ 1_Skeleton-based Action Recognition Using LSTM and CNN https://tryolabs.com/blog/2017/08/30/object-detection-an-overview-in-the-age-of-deep-learning/
https://zhuanlan.zhihu.com/p/27657196
https://www.oreilly.com/ideas/visualizing-convolutional-neural-networks
https://zhuanlan.zhihu.com/p/29435406
视频行为识别：https://zhuanlan.zhihu.com/wzmsltw
用pandas处理大数据——节省90%内存消耗的小贴士
Minimizing the Negative Log-Likelihood： http://willwolf.io/2017/05/18/minimizing_the_negative_log_likelihood_in_english/