Anti-cheating method and system for image annotation and electronic equipment

Abstract

The invention provides an anti-cheating method for image annotation. The anti-cheating method comprises the steps: obtaining at least one to-be-annotated image with a to-be-annotated object; and obtaining the frame selection time of the current operation and the to-be-annotated object selected by the previous operation for at least one time and the annotation time of the to-be-annotated object, judging whether the frame selection time and the annotation time of the current operation and the previous operation for at least one time exceed preset thresholds or not, and judging whether the operation is cheating or not based on a judgment result. By judging whether two times of frame selection time and annotation time exceed a preset threshold value or not, the method anti-cheating determineswhether cheating occurs is determined, and improves the constraint force for employee operation, so as to avoid the problems that the employee repeatedly carries out picture frame and repeatedly selects picture annotation on the same to-be-annotated object or characteristic, thus guaranteeing annotation quality of image annotation, and reducing time and investment cost caused by cheating problems.

Description

【Technical field】 [0001] The invention relates to the field of image processing, in particular to an anti-cheating method, system and electronic equipment for image labeling. 【Background technique】 [0002] In the field of image annotation, it is necessary to annotate the image and then input it into the neural network for supervised training of the neural network. However, the convergence of the neural network needs to train a large number of labeled images, which makes the image labeling work extremely heavy. [0003] Existing companies usually outsource the task of labeling images and introduce external personnel to process the data. However, outsourced employees cheated when labeling images, that is, after repeatedly labeling the same feature, they obtained the same multiple training samples, thereby increasing the number of training samples to increase income, and reducing image quality and neural network models. training efficiency. [0004] In the existing anti-che...

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the problem of low efficiency in the current anti-cheating method of image annotation, the present invention provides an anti-cheating method of image annotation

Images