项目介绍

python离线CNN车牌识别源码+演讲稿

系统说明

Hello, classmates. Today I would like to use CNN realize to License Plate Recognition. My code is based on TensorFlow, a deep learning framework

First of all, let’s see what’s CNN and why we use CNN. The neurons in each layer of a CNN are convolution kernels, which is different from normal neurons. As for the advantages of CNN are as follows. CNN can get the spatial information of the image, solve the problem of too many parameters, low efficiency and difficult training, and reduce the possibility of over-fitting.

The task of license plate recognition is divided into three parts: license plate location, license plate screening, character recognition.

Firstly, the image is preprocessed and get binarization image by using the threshold. Secondly, I screen the license plate location by contour features and overflowing water filling algorithm. Further more, The CNN is trained to classify the license plate and non-license plate images so as to screen out the real license plate.

And then, we have known that license plate include seven characters, so we can easily divide each characters. Each character recognition is treated as a multi-classification problem by using CNN. And after recognition, I output and display the identification results.

Thanks for your listening.

适用场景：

毕业论文、课程设计、公司项目参考

运行截图

关注【程序代做 源码分享】公众号获取更多免费源码！！！