- 1
- 2
- 3
- 4
- 5
CNN深度学习的验证码识别及Android平台移植
资料介绍
针对目前网络生活应用中越来越多的图片验证码,提出了一种基于卷积神经网络的验证码识别方案,使用4 500张四位数字验证码作为训练集,利用TensorFlow深度学习框架训练网络模型,经过测试集验证后正确率为98.55%。网络验证通过后将该含参数的网络模型移植至Android平台,采用OpenCV对验证码图片进行预处理,并利用该网络模型实现验证码识别。
Aiming at CAPTCHA identification problems in current network life,a CAPTCHA identification scheme based on convolutional neural network is proposed.Using 4 500 digital CAPTCHA as training sets,TensorFlow deep learning framework is used to train network models.After the test set verification,the correct rate is 98.55%.After the network verification is passed,the parameter-containing network model is transplanted to the Android platform,and CAPTCHA is preprocessed by OpenCV,and the CAPTCHA identification is realized by using the network model.
部分文件列表
文件名 | 大小 |
CNN深度学习的验证码识别及Android平台移植.pdf | 1M |
全部评论(0)