1results about How to "More attention" patented technology

The application discloses a face segmentation model training method and a face image segmentation method. The method comprises the following steps: obtaining a training sample set; the training sample set comprises a plurality of sample images; inputting the training sample set into a face segmentation model for encoding processing and decoding processing to obtain a plurality of predicted images; performing Hungarian matching on at least one to-be-segmented face region marked in the plurality of sample images and at least one predicted face region included in the plurality of predicted images to obtain at least one unmatched face region; determining a global loss value based on the plurality of sample images and the plurality of predicted images; determining a local loss value based on the regions of the at least one unmatched face region in the plurality of sample images and the plurality of predicted images; determining a target loss value based on the global loss value and the local loss value; adjusting the parameters of the face segmentation model based on the target loss value until a preset training condition is reached to obtain a trained face segmentation model.