GitPlanet

Cheap and reliable Node.js hosting starts at $3/month, and $1/month static HTML hosting

Created with love in Canada, visit hostnodejs.com today

Feel like to post an Ad? Learn Details
All Projects → taosir → Cnn_handwritten_chinese_recognition

taosir / Cnn_handwritten_chinese_recognition

Licence: mit
CNN在线识别手写中文。

Programming Languages

python
139335 projects - #7 most used programming language

Labels

flaskcnnchineserecognition

Projects that are alternatives of or similar to Cnn handwritten chinese recognition

Pytorch classification
利用pytorch实现图像分类的一个完整的代码，训练，预测，TTA，模型融合，模型部署，cnn提取特征，svm或者随机森林等进行分类，模型蒸馏，一个完整的代码
Stars: ✭ 395 (+8.22%)
Mutual labels:  flask, cnn
3d Densenet
3D Dense Connected Convolutional Network (3D-DenseNet for action recognition)
Stars: ✭ 118 (-67.67%)
Mutual labels:  cnn, recognition
Food Recipe Cnn
food image to recipe with deep convolutional neural networks.
Stars: ✭ 448 (+22.74%)
Mutual labels:  cnn, recognition
Chinese Text Classification
Chinese-Text-Classification，Tensorflow CNN（卷积神经网络）实现的中文文本分类。QQ群：522785813，微信群二维码：http://www.tensorflownews.com/
Stars: ✭ 284 (-22.19%)
Mutual labels:  chinese, cnn
Cnn Question Classification Keras
Chinese Question Classifier (Keras Implementation) on BQuLD
Stars: ✭ 28 (-92.33%)
Mutual labels:  chinese, cnn
Trape
People tracker on the Internet: OSINT analysis and research tool by Jose Pino
Stars: ✭ 6,753 (+1750.14%)
Mutual labels:  flask, recognition
Dialectid e2e
End to End Dialect Identification using Convolutional Neural Network
Stars: ✭ 40 (-89.04%)
Mutual labels:  cnn, recognition
Phormatics
Using A.I. and computer vision to build a virtual personal fitness trainer. (Most Startup-Viable Hack - HackNYU2018)
Stars: ✭ 79 (-78.36%)
Mutual labels:  flask, recognition
Cnn Text Classification Tf Chinese
CNN for Chinese Text Classification in Tensorflow
Stars: ✭ 237 (-35.07%)
Mutual labels:  chinese, cnn
Text Classification Cnn Rnn
CNN-RNN中文文本分类，基于TensorFlow
Stars: ✭ 3,613 (+889.86%)
Mutual labels:  chinese, cnn
Flask Sqlalchemy
Adds SQLAlchemy support to Flask
Stars: ✭ 3,658 (+902.19%)
Mutual labels:  flask
Cnn lstm ctc tensorflow
CNN+LSTM+CTC based OCR implemented using tensorflow.
Stars: ✭ 343 (-6.03%)
Mutual labels:  cnn
Time Series Prediction
A collection of time series prediction methods: rnn, seq2seq, cnn, wavenet, transformer, unet, n-beats, gan, kalman-filter
Stars: ✭ 351 (-3.84%)
Mutual labels:  cnn
Pytest Flask
A set of pytest fixtures to test Flask applications
Stars: ✭ 359 (-1.64%)
Mutual labels:  flask
Numpy neural network
仅使用numpy从头开始实现神经网络,包括反向传播公式推导过程; numpy构建全连接层、卷积层、池化层、Flatten层；以及图像分类案例及精调网络案例等,持续更新中... ...
Stars: ✭ 339 (-7.12%)
Mutual labels:  cnn
Microblog
A microblogging web application written in Python and Flask that I developed as part of my Flask Mega-Tutorial series.
Stars: ✭ 3,788 (+937.81%)
Mutual labels:  flask
Personality Detection
Implementation of a hierarchical CNN based model to detect Big Five personality traits
Stars: ✭ 338 (-7.4%)
Mutual labels:  cnn
Cmdb
CMDB: configuration and management of IT resources, demo: http://121.42.12.46:8000
Stars: ✭ 337 (-7.67%)
Mutual labels:  flask
Meinheld Gunicorn Flask Docker
Docker image with Meinheld and Gunicorn for Flask applications in Python. Optionally with Alpine Linux.
Stars: ✭ 336 (-7.95%)
Mutual labels:  flask
Cope
A modern IDE for writing classical Chinese poetry 格律诗编辑程序
Stars: ✭ 362 (-0.82%)
Mutual labels:  chinese
View All Similar Projects ➔

cnn_handwritten_chinese_recognition

     使用python+flask搭建的一个网站，然后从网页的写字板上获取鼠标手写的汉字经过转码后传回后台，并经过图片裁剪处理之后传入CNN手写中文识别的模型中进行识别，最后通过PIL将识别结果生成图片，最后异步回传给web端进行识别结果展示。中文总共50,000多汉字，常用的有3,755个。这里主要对常见的3755个汉字进行识别。
demogif

一、数据集

     目前国内有很多优秀的中文手写识别数据集。例如：北京邮电大学模式识别实验室发布的数据(HCL2000)，它是目前最大的脱机手写汉字库，共有1,000个人书写，除了汉字样本库外，还有一个对应的书写者信息库，记录了书写者的年龄、职业、文化程度等信息，用于研究相关影响因素。目前此数据库免费向研究者公开。本文使用的是中科院自动研究所的分享的中文手写数据集CASIA-HWDB(下载地址http://www.nlpr.ia.ac.cn/databases/handwriting/Home.html )，由187个人通过Wacom笔在线输入的手写汉字。

二、CNN结构：

     用tensorflow库来实现【三个卷积层+三个池化层+两个全连接层】的卷积神经网络，结构如下图所示：
cnn_model_arch      训练图片都修整成64x64大小，这里我只训练常见的3755个汉字，在CNN识别数字的模型结构上再添加了一个卷积层和池化层，其他结构差不多。

     将下载好的HWDB数据集解压处理好开始训练，这个训练过程比较长，我最终在GPU:GTX1050Ti上迭代了12,000次花费几个小时，最终取最可能的前三个预测值

三、加载模型

     我训练模型迭代了12,000次之后，将训练参数保存在checkpoint文件夹中，不过因为单个文件大小的限制，训练好的模型文件从百度云上下载：链接: https://pan.baidu.com/s/1eSWBIyI 密码: kv2r;

     下载之后直接覆盖checkpoint文件夹。

四、环境

  • python 3.6.1;
  • pipenv *;
  • flask 0.12.2;
  • tensorflow 1.3.0;
  • pillow 4.2.1;
  • pickleshare 0.7.4;
  • numpy 1.13.1;

五、运行

1、克隆项目，先安装python3.0和pipenv，然后 pipenv install --three；
2、从百度云下载训练好的模型文件，放到相应的checkpoint文件夹下；
3、使用pipenv run python3 run.py运行；
4、打开本地浏览器输入localhost:5000进行查看；

六、其他

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].