Action Recognition Using Undecimated Dual Tree Complex Wavelet Transform From Depth Motion Maps / Depth Sequences

Abstract

<p><strong>Abstract.</strong> Accumulating the motion information from a video sequence is one of the highly challenging and significant phase in Human Action Recognition. To achieve this, several classical and compact representations are proposed by the research community with proven applicability. In this paper, we propose a compact Depth Motion Map based representation methodology with hastey striding, consisely accumulating the motion information. We extract Undecimated Dual Tree Complex Wavelet Transform features from the proposed DMM, to form an efficient feature descriptor. We designate a Sequential Extreme Learning Machine for classifying the human action secquences on benchmark datasets, MSR Action 3D dataset and DHA Dataset. We empirically prove the feasability of our method under standard protocols, achieving proven results.</p>

Complex Gradient Function Based Descriptor for Iris Biometrics and Action Recognition

“…The CGF descriptor based methodology achieved significantly better results than many of the existing methods available in the literature. Our experiments on MSR Action3D shows an average accuracy of 96.39% DMMHOG [27] 85.5 ROP [24] 86.5 HON4D [17] 88.9 DMM-LBP [4] 91.9 SDMM-UDTCWT [20] 93.41 Proposed 95. 24 under setting1, as presented in Table 4 whereas, Table 5 presents the cross subject test results on the same dataset under setting2 demonstrating significant improvements compared to the available results.…”

“…24 under setting1, as presented in Table 4 whereas, Table 5 presents the cross subject test results on the same dataset under setting2 demonstrating significant improvements compared to the available results. Referring back to the works such as [20,27], it is observant that interclass similarity among the samples belonging to Draw tick, Draw x and Draw circle is very high, also the samples belonging to these classes are mostly misclassified in the above cited existing works. In our experimentation, it is evident that six samples belonging to Draw x are misclassified as Draw tick class.…”

“…We demonstrate a detailed comparative study of our results with the results available in the published literatures, under the same train/test protocols used in [20,22]. The MSR Action 3D dataset is treated under two different settings for a justifiable experimentation, namely Setting1 and Setting2.…”

“…As a preprocessing stage, we construct a compact and dense representation namely Stridden Depth Motion Map (SDMM) [20,22] by piling up the action cues from the depth frames of the depth video sequence. This is achieved by traversing consecutive depth frames in the video in steps of four frames at a time and totaling up the absolute value of differences from one frame to its second predecessor frame.…”

Exploring 3D Human Action Recognition Using STACOG on Multi-View Depth Motion Maps Sequences

“…They used the fisher kernel representation method to fuse obtained features and the KLM classifier to detect action. After that, Shekar et al [ 35 ] introduced Stridden DMMs from which effective information of actions can be obtained quickly. They Undecimated the Dual-Tree Complex Wavelet Transform algorithm to extract wavelet (UDTCWT) features from the proposed DMMs.…”