| 摘 要: 目前许多网站使用验证码阻止黑客进行暴力破解登录口令，但是随着技术的发展验证码识别的难度及 成本日益提升，传统的OCR(Optical Character Recognition)技术识别效率已不能满足需求。Vcode Identify平台使用 TensorFlow建立深度学习模型并结合Web应用开发的一款验证码识别软件，使用该平台用户不仅可以通过调用接口或 者上传文件的形式对验证码进行识别，而且还可以建立新模型并训练，进而可以识别新类型验证码，该软件使用简单、 扩展性强。 |
| 关键词: 深度学习；Python Web；验证码识别；Vcode Identify平台 |
|
中图分类号: TP315
文献标识码: A
|
| 基金项目: 江苏省大学生创新创业训练项目(项目编号：xcx2019076). |
|
| Web Application Platform for Captcha Breaking Based on Deep Learning |
|
WANG Hao,KANG Xiaofeng,LU Zhike,SHI Runjie,HUANG Chengxin1,2,3,4,5
|
1.( College of Information Engineering, Xuzhou Institute of Technology, Xuzhou 221000, China) vmzkmerq@gmail.com;2.kxfeng07@163.com;306948038@qq.com;4.handchongboy@sina.com;5.761369543@qq.com
|
| Abstract: At present,many websites use verification codes to prevent hackers from performing brute-forced login passwords.However,the re-adoption of technology to develop verification codes increases the complexity and cost of recognition.The traditional OCR (Optical Character Recognition) technology recognition has been unable to meet efficiency needs.The Vcode Identify recognition platform uses TensorFlow to build a deep learning model.The verification code recognition software was developed in conjunction with a web application.Using this platform,users not only can identify verification codes by calling interfaces or uploading files but also can build and train new models to identify the new type of verification code.This system is simple to use and highly scalable. |
| Keywords: deep learning;python web;captcha breaking;Vcode Identify platform |
