Face reconstruction method based on learning

The present disclosure provides a computer-implemented method of performing machine learning on a facial image. The method is performed by one or more computing devices and includes: obtaining a training data set, the training data set including a face image and a first face modeling parameter set associated with a three-dimensional model of the face; inputting the face image to a face reconstruction system, the face reconstruction system comprising one or more machine-learned neural networks; receiving a second face modeling parameter set generated by the face reconstruction system based at least in part on the face image; evaluating a loss function comparing the first face modeling parameter set and the second face modeling parameter set; and training one or more machine-learned convolutional neural networks of the facial reconstruction system based at least in part on the loss function. 本公开提供了一种在面部图像上执行机器学习的计算机实现的方法。该方法由一个或多个计算设备执行，并包括：获取训练数据集，该训练数据集包括面部图像和与面部的三维模型相关联的第一面部建模参数集；将面部图像输入到面部重建系统，该面部重建系统包括一个或多