Method and device for information generation

The invention discloses a method and device for information generation. The method comprises steps that a sequence of images acquired through photographing an infant is acquired; the images of the image sequence are sequentially inputted to a pre-trained first region positioning model to acquire the skeleton image region sequence of the infant; based on the skeleton image region sequence of the infant and a pre-trained first characteristic extraction model, the skeleton characteristic sequence of the infant is acquired; and the skeleton characteristic sequence of the infant is inputted into apre-trained mental state prediction model to acquire the mental state information of the infant. The method is advantaged in that prediction accuracy of the mental state information of the infant is improved. 本申请实施例公开了用于生成信息的方法和装置。该方法的具体实施方式包括：获取对婴儿进行拍摄所得到的图像序列；将图像序列中的图像依次输入至预先训练的第区域定位模型，得到婴儿的骨架图像区域序列；基于所述婴儿的骨架图像区域序列和预先训练的第特征提取模型，得到所述婴儿的骨架特征序列；将婴儿的骨架特征序列输入至预先训练的心理状态预测模型，得到婴儿的心理状态信息。该实施方式提高了对婴儿的心理状态信息的预测准确度。