An Efficient Approach for Multi-view Human Action Recognition Based on Bag-of-Key-Poses

Fusion of Skeletal and Silhouette-Based Features for Human Action Recognition with RGB-D Devices

“…As has been shown in [5,6], this learning method handles multiple views successfully. In the present work, the skeleton and the silhouette could be considered as different views.…”

“…As learning algorithm we employ a method based on bag of key poses [5,6]. Similar to the bag-of-words-model, first, a codebook -called bag of key poses -is obtained using the -means clustering algorithm.…”

Evolutionary joint selection to improve human action recognition with RGB-D devices

“…In order to evaluate a specific feature subset selection and to obtain the fitness value of the corresponding individual, a human action recognition method based on the one from Chaaraoui et al (2012) is proposed. An overview of the method can be seen in Fig.…”

“…Regarding the feature extraction process, human silhouettes are used in Chaaraoui et al (2012). Translating this to the domain of skeletal data, a In this sense, position, scale and rotation invariance is achieved by applying the proposed normalisation process detailed in Algorithm 2.…”

“…(L1-norm) (Azary & Savakis, 2012) 77.66% 73.17% 91.58% 80.80% Sparse Repr. (L2-norm) (Azary & Savakis, 2012) 76.60% 75.61% 89.47% 80.56% Keyposes and Decision Forests (Miranda et al, 2012) 93.5% 52% 95.4% 80.30% Histograms of 3D Joints (Xia et al, 2012) 87.98% 85.48% 63.46% 78.97% Bag of 3D Points (Li et al, 2010) 72.9% 71.9% 79.2% 74.67% Bag of key poses and DTW (Chaaraoui et al, 2012) 78 The size of the population has been set to N = 10 individuals. Instead of having a static mutation probability, each time a random value for p mut in the interval [0, 0.2] is selected.…”