Image processing model training method and device, storage medium and electronic equipment

Abstract

The invention discloses an image processing model training method and device, a storage medium and electronic equipment. The method comprises the steps of obtaining a sample loss value obtained after simulation training is performed on each training sample in each iteration period; according to a plurality of sample loss values obtained by the obtained training sample in a plurality of iteration periods, determining a sample type to which the training sample belongs, the sample type being used for indicating a convergence degree of the training sample reaching a convergence condition in simulation training; and adding an attention mark to a target training sample of the target sample type, and carrying out model training on the original model by using the target training sample until a target model reaching a second convergence condition matched with the target sample type is obtained, the attention mark is used for indicating to increase the training weight of the target training sample in the model training in each iteration period. The technical problem of poor image processing model performance caused by improper model training is solved.

Description

Technical field [0001] The present invention relates to the field of computer, and in particular, a training method and apparatus, a storage medium, and an electronic device of an image processing model. Background technique [0002] Currently, the image processing model is increasingly popular in data processing. The accuracy of the data processing in the application is usually related to model training in the application. In the model training phase, training data has great impact on model performance. [0003] In the model training, it is usually only trained with the positive sample and negative samples, and there is no further classification of the sample image. For sample images of different training difficulty, the model is preded during the model training, which can reduce the performance of the image processing model of training due to the influence of noise. [0004] For the above problems, an effective solution has not yet been proposed. Inventive content [0005] Emb...

AI Technical Summary

Problems solved by technology

[0005] Embodiments of the present invention provide a training method and device for an image processing model, a storage medium, and electronic equipment, so as to at least solve the technical problem of poor performance of the image processing model caused by improper model training

Images