VR action evaluation method and system based on graph convolution deep learning model

The invention discloses a VR action evaluation method and system based on a graph convolution deep learning model. The method comprises the following steps: acquiring an image data set according to collected action data in a virtual environment; inputting an image convolution kernel of the image data set into an image attention module, and determining collected specific action data; training the collected specific action data set and the general action data set to obtain a graph convolution deep learning VR action evaluation model; and according to the graph convolution deep learning VR action evaluation model, evaluating the electric power operation field action. Therefore, the hand action of the power operation training in the virtual environment is detected by training the space-time diagram convolutional neural network, and the accuracy of action evaluation is improved. 本发明公开了一种基于图卷积深度学习模型的VR动作评价方法及系统。其中，该方法包括：包括：根据采集到的虚拟环境中的动作数据，获得图像数据集；将所述图像数据集的图卷积核输入到图注意力模块中，确定采集的特定动作数据；对所述采集的特定动作数据集和通用动作数据集进行训练，获得图卷积深