Network verification code recognition method and device based on deep learning and computer equipment

Abstract

The invention relates to a network verification code recognition method and device based on deep learning, and computer equipment. The network verification code recognition method comprises the steps:obtaining to-be-recognized verification code data; inputting the verification code data to be recognized into the verification code recognition model for character recognition to obtain a recognitionresult; judging whether the identification result is correct; if the identification result is correct, automatically labeling the corresponding verification code data to enter a training set and a test set; and if the identification result is incorrect, re-labeling the verification code data, and putting the automatically labeled verification code data and the re-labeled verification code data into the verification code identification model for re-training to update the verification code identification model until the identification rate of the verification code identification model reaches apreset threshold. By taking a small amount of verification code data as a verification code sample for training, and updating the verification code recognition model according to a training result, the network verification code recognition method can effectively reduce annotation samples required for training / learning of the verification code recognition model, and meanwhile, can improve the model recognition precision.

Description

technical field [0001] The present invention relates to the field of machine learning, and more specifically refers to a network verification code recognition method, device and computer equipment based on deep learning. Background technique [0002] CAPTCHA is the abbreviation of "Completely Automated Public Turing test totellComputers and Humans Apart" (Turing test to tell computers and humans fully automatically), which is a public fully automatic program to distinguish whether the user is a computer or a human. Research on verification code recognition technology is the category of artificial intelligence. Through the research on verification code recognition technology, the machine can achieve the recognition effect of human eyes, which has a great role in promoting the development of artificial intelligence. At the same time, scientific research institutions and researchers engaged in big data analysis need to obtain a large amount of data on the Internet for scientifi...

AI Technical Summary

Problems solved by technology

For network verification codes, the most common four-digit letter + number combination, there are a total of 36^4 = 1679616 combinations. For twisted and glued verification codes that cannot be cut, a combination requires at least 10 pictures, so all samples need at least 1679616* 10 = 16796160 pictures, such a labeling workload is huge, and using a manual coding platform also requires a lot of cost

If the sample size is too small, it will greatly affect the accuracy of model recognition

Images