Tai Chi Movement Recognition Method Based on Deep Learning Algorithm

Abstract

The current action recognition method has good effect when applied to static recognition, but, when applied to dynamic action sequence recognition, the temporal and spatial feature segmentation is too dependent on sample template, resulting in low recognition accuracy. To address the inadequacies of standard movement detection techniques in the application of comparable domains, a deep learning algorithm is utilised to recognise Tai Chi Chuan motions. For Tai Chi Chuan movement human body skeleton framework, add image depth parameter is added, and OpenPose model is utilised to estimate joint point coordinates. The ST-GCN deep learning model was created to recognise Tai Chi Chuan motions by extracting movement features from the spatiotemporal trajectory of human joints during Tai Chi Chuan movements. Instance test results show that rate of using the deep learning algorithm of gesture recognition is 89.22%, with significantly lower error detection rate, which is good for Tai chi chuan movement recognition effect.

Landscape Design of Rural Characteristic Towns Based on Big Data Technology

“…e proposed method is more efficient methods than deep learning algorithms. When the cloud point data registration results of the deep learning algorithm method [26][27][28][29] and the proposed method are obtained, and the cloud point data are continuously improved, the convergence of the proposed method's cloud point data registration results is lower than that of the deep learning algorithm method. It is proved that the convergence of the proposed method is better than that of the deep learning algorithm method.…”

Improving Small-Scale Human Action Recognition Performance Using a 3D Heatmap Volume

“…Dong et al [ 65 ] also proposed a Tai Chi dataset called ‘Sub-Tai chi’ consisting of 15 actions and applied structural LSTM with an attention module for recognition; they reached 79% recognition accuracy on their own dataset. Liu et al [ 66 ] applied the ST-GCN model on their own Tai Chi dataset and achieved 89.22% recognition accuracy. All these studies created their own Tai Chi datasets and are not available on websites for comparison.…”