Method and system for increasing difficult sample data, computer equipment and storage medium

Abstract

The invention discloses a method and system for increasing difficult sample data, computer equipment and a storage medium, and the method comprises the steps: obtaining an original training set with a plurality of simple sample images and difficult sample images, training an initial segmentation model through the original training set, and obtaining a target segmentation model; screening the original training set to obtain a difficult sample image, and training a target segmentation model by using the difficult sample image to obtain an optimized target segmentation model; inputting the target image into the optimized target segmentation model for identification, and judging whether the identification result is a difficult sample image; and if the identification result is a difficult sample image, storing the target image as a new difficult sample image in the original training set. According to the method, the target segmentation model for identifying the difficult sample data is obtained by training the conventional initial segmentation model, so that the rapid identification of the difficult sample data is realized, the identification efficiency is improved, and the acquisition cost is reduced.

Description

technical field [0001] The present invention relates to the technical field of data processing, and in particular, to a method, a system, a computer device and a storage medium for adding difficult sample data. Background technique [0002] With the rapid development of video special effects, the segmentation of various instances has gradually entered the public eye. At present, the segmentation methods of various instances are mainly based on semantic and image matting. Both methods have good performance for conventional scenes, but in practical applications The problem of scarcity of hard samples is often encountered. [0003] For the scarcity of difficult samples, there are currently two main solutions. One is to modify the model structure and add detail processing branches to solve the problem of poor segmentation of a certain type of samples. The other is to improve the accuracy of the model on difficult samples by increasing the number of data annotations for difficul...

AI Technical Summary

Problems solved by technology

[0004] Embodiments of the present invention provide methods, systems, computer equipment, and storage media for increasing difficult sample data, aiming to solve the problems of high cost and long time-consuming acquisition of difficult sample data in the prior art

Images