Human face expression image processing method and device based on monocular RGB camera

The embodiment of the invention discloses a face expression image processing method and device based on a monocular RGB camera, and the method comprises the steps: obtaining an RGB video, and extracting a face key point group of a target video frame in the RGB video, the face key point group comprising a plurality of target key points; performing data preprocessing on the face key points to obtain a processed face key point group; the processed face key point group is input into a pre-trained face capture driving model to obtain expression driving data, and the expression driving data is used for being sent to an application end and driving a virtual character; wherein the face capture driving model is obtained by training face key point group samples based on a deep learning network, and the face key point group samples are obtained by performing data preprocessing on key points extracted from RGB video samples. The technical problems that in the prior art, facial expression driving is poor in compatibility