3results about How to "Simplify the data collection process" patented technology

The present disclosure provides a reconstruction method and device of a human virtual avatar, a terminal and a storage medium. The reconstruction method of the human virtual avatar comprises: acquiring a plurality of images including a human body and a virtual human data set; acquiring a depth map from the plurality of images; and performing three-dimensional reconstruction and rendering based on the plurality of images, the depth map and the virtual human data set by using a trained neural network model to obtain a human three-dimensional model and a corresponding rendered image. The present disclosure can reconstruct the human virtual avatar by using a small number of images, thereby simplifying the data acquisition process. In addition, the reconstruction method of the present disclosure can ensure high calculation efficiency while ensuring rendering quality, and can be used for real-time interaction.

The invention discloses a smart classroom emotion recognition method and system based on dynamic facial expressions. The method comprises the following steps: firstly, acquiring a classroom panoramic video stream, and preprocessing the classroom panoramic video stream to generate a structured facial image sequence; the method comprises the following steps of: obtaining a fused video frame through a space-time attention fusion module, then embedding time and space tokens, inputting the fused video frame into a double-scanning bidirectional Mamaba block to carry out space-time feature extraction and modeling, finally outputting the probability of each emotion through a pre-trained emotion classification model, and selecting the emotion with the highest probability as final output to be fed back to a user. The dynamic facial expression recognition system provided by the invention is light in weight and high in performance, so that facial expressions of students can be accurately recognized in an actual classroom, and the reliability and the stability of results are ensured. The method solves the problems of weak dynamic modeling, strong multi-mode dependence, lack of education closed loop and the like in a classroom scene in the prior art, and is suitable for non-sentible emotion perception and teaching optimization in intelligent education.

This invention discloses a method for constructing a structured light dataset based on a digital twin model of a fringe projection system. A digital twin model of the system is constructed in computer graphics software based on the hardware and calibration parameters of a real fringe projection system. Then, a node tree is used to simulate the surface properties of the object under test and the system's ambient light source is set. The parameters are adjusted to diversify the target surface properties and ambient light. Fringe images are then imported, and automated projection and data acquisition of the digital twin model are achieved through programming. Finally, fringe images and corresponding depth images that meet the user's quantity requirements are generated, completing the construction of the structured light dataset. This invention solves the problems of complex fringe projection data acquisition processes, high dataset production costs, and limited data samples in real-world scenarios, achieving rapid and high-precision construction of fringe projection system datasets.