STFC: Spatio-temporal feature chain for skeleton-based human action recognition

Human action recognition via multi-task learning base on spatial–temporal feature

“…3 Laptev et al [26] 91.8 Le et al [29] 93. 9 Wang et al [46] 92.1 Ballan et al [2] 92. 7 Zhang and Tao [54] 93.2 Yang et al [50] 93.…”

“…This method shows very good performance in web video concept detection. Ding et al [9] applied some new technologies and proposed a method to recognize human actions form sequences of 3D joint positions. The main contribution of this paper is that author proposed the spatio-temporal feature chain that is introduced to represented the characteristic parameters of temporal sequential patterns.…”

HIF3D: Handwriting-Inspired Features for 3D skeleton-based action recognition

“…LTI + HMM, 2014 [18] 86.76 Grassmann + SVM, 2015 [19] 88.5 HOJ3D + HMM, 2012 [6] 90.95 STFC + SVM, 2015 [8] 91.5…”

“…Approaches of the first category measure the similarity between the action to be recognized and a stored template while taking into account all allowable distortions. Several works followed this trend among which we can find the Spatio Temporal Feature Chain (STFC) [8] and the Linear Dynamical Systems (LDS) [9]. In the statistical approaches, each action is represented in terms of d features or measurements and is viewed as a point in a d-dimensional space.…”

A Comparative Study on Human Action Recognition Using Multiple Skeletal Features and Multiclass Support Vector Machine

“…They used this representation to select the most informative frames from a large pool of action sequences. Ding et al [21] proposed a method to learn the low dimensional embedding with a manifold functional variant of principal component analysis (mfPCA). Fletcher et al [22] developed a method of principal geodesic analysis, a generalization of principal component analysis to the manifold setting.…”