push all files

.gitignore

@@ -0,0 +1,107 @@
+modelAngle.h5
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+env/
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+.hypothesis/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# pyenv
+.python-version
+
+# celery beat schedule file
+celerybeat-schedule
+
+# SageMath parsed files
+*.sage.py
+
+# dotenv
+.env
+
+# virtualenv
+.venv
+venv/
+ENV/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+
+.DS_Store
+.idea/
+*.h5

.vscode/launch.json

@@ -0,0 +1,116 @@
+{
+    // 使用 IntelliSense 了解相关属性。 
+    // 悬停以查看现有属性的描述。
+    // 欲了解更多信息，请访问: https://go.microsoft.com/fwlink/?linkid=830387
+    "version": "0.2.0",
+    "configurations": [
+        {
+            "name": "Python: Current File",
+            "type": "python",
+            "request": "launch",
+            "program": "${file}"
+        },
+        {
+            "name": "Python: Attach",
+            "type": "python",
+            "request": "attach",
+            "localRoot": "${workspaceFolder}",
+            "remoteRoot": "${workspaceFolder}",
+            "port": 3000,
+            "secret": "my_secret",
+            "host": "localhost"
+        },
+        {
+            "name": "Python: Terminal (integrated)",
+            "type": "python",
+            "request": "launch",
+            "program": "${file}",
+            "console": "integratedTerminal"
+        },
+        {
+            "name": "Python: Terminal (external)",
+            "type": "python",
+            "request": "launch",
+            "program": "${file}",
+            "console": "externalTerminal"
+        },
+        {
+            "name": "Python: Django",
+            "type": "python",
+            "request": "launch",
+            "program": "${workspaceFolder}/manage.py",
+            "args": [
+                "runserver",
+                "--noreload",
+                "--nothreading"
+            ],
+            "debugOptions": [
+                "RedirectOutput",
+                "Django"
+            ]
+        },
+        {
+            "name": "Python: Flask (0.11.x or later)",
+            "type": "python",
+            "request": "launch",
+            "module": "flask",
+            "env": {
+                "FLASK_APP": "${workspaceFolder}/app.py"
+            },
+            "args": [
+                "run",
+                "--no-debugger",
+                "--no-reload"
+            ]
+        },
+        {
+            "name": "Python: Module",
+            "type": "python",
+            "request": "launch",
+            "module": "module.name"
+        },
+        {
+            "name": "Python: Pyramid",
+            "type": "python",
+            "request": "launch",
+            "args": [
+                "${workspaceFolder}/development.ini"
+            ],
+            "debugOptions": [
+                "RedirectOutput",
+                "Pyramid"
+            ]
+        },
+        {
+            "name": "Python: Watson",
+            "type": "python",
+            "request": "launch",
+            "program": "${workspaceFolder}/console.py",
+            "args": [
+                "dev",
+                "runserver",
+                "--noreload=True"
+            ]
+        },
+        {
+            "name": "Python: All debug Options",
+            "type": "python",
+            "request": "launch",
+            "pythonPath": "${config:python.pythonPath}",
+            "program": "${file}",
+            "module": "module.name",
+            "env": {
+                "VAR1": "1",
+                "VAR2": "2"
+            },
+            "envFile": "${workspaceFolder}/.env",
+            "args": [
+                "arg1",
+                "arg2"
+            ],
+            "debugOptions": [
+                "RedirectOutput"
+            ]
+        }
+    ]
+}

.vscode/settings.json

@@ -0,0 +1,4 @@
+{
+    "python.pythonPath": "/anaconda3/bin/python",
+    "git.ignoreLimitWarning": true
+}

README.md

@@ -0,0 +1,76 @@
+# 本文基于tensorflow、keras/pytorch实现对自然场景的文字检测及端到端的OCR中文文字识别
+
+# 实现功能
+
+- [x]  文字方向检测 0、90、180、270度检测 
+- [x] 文字检测 后期将切换到keras版本文本检测 实现keras端到端的文本检测及识别
+- [x] 不定长OCR识别
+- [x] 增加python3.6 支持
+
+
+## 环境部署
+``` Bash
+##GPU环境
+sh setup.sh
+##CPU环境
+sh setup-cpu.sh
+##CPU python3环境
+sh setup-python3.sh
+```
+
+# 模型训练
+
+## 训练keras版本的crnn   
+
+``` Bash
+cd train & sh train-keras.sh   
+```
+
+## 训练pytorch版本的crnn   
+
+``` Bash
+cd train & sh train-pytorch.sh   
+```
+# 文字方向检测
+基于图像分类，在VGG16模型的基础上，训练0、90、180、270度检测的分类模型，详细代码参考angle/predict.py文件，训练图片8000张，准确率88.23%。
+模型地址[百度云](https://pan.baidu.com/s/1pM2ha5P)下载
+
+# 文字检测
+支持CPU、GPU环境，一键部署，
+[文本检测训练参考](https://github.com/eragonruan/text-detection-ctpn)  
+
+
+# OCR 端到端识别:GRU+CTC
+## ocr识别采用GRU+CTC端到到识别技术，实现不分隔识别不定长文字
+提供keras 与pytorch版本的训练代码，在理解keras的基础上，可以切换到pytorch版本，此版本更稳定
+- 此外还添加了tensorflow版本的资源仓库：[TF:LSTM-CTC_loss](https://github.com/ilovin/lstm_ctc_ocr)    
+
+
+# 识别结果展示
+## 文字检测及OCR识别结果
+<div>
+<img width="300" height="300" src="https://github.com/chineseocr/chinses-ocr/blob/master/img/tmp.jpg"/>
+<img width="300" height="300" src="https://github.com/chineseocr/chinses-ocr/blob/master/img/tmp.png"/>
+</div>
+
+### 倾斜文字 
+
+<div>
+<img width="300" height="300" src="https://github.com/chineseocr/chinses-ocr/blob/master/img/tmp1.jpg"/>
+<img width="300" height="300" src="https://github.com/chineseocr/chinses-ocr/blob/master/img/tmp1.png"/>
+</div>
+
+## 参考
+```
+1.crnn 
+https://github.com/meijieru/crnn.pytorch.git      
+2.keras-crnn 版本实现参考 https://www.zhihu.com/question/59645822  
+
+3.tensorflow-crnn 
+https://github.com/ilovin/lstm_ctc_ocr      
+
+3.ctpn
+https://github.com/eragonruan/text-detection-ctpn 
+https://github.com/tianzhi0549/CTPN   
+```
+

angle/__init__.py

@@ -0,0 +1,6 @@
+#!/usr/bin/env python2
+# -*- coding: utf-8 -*-
+"""
+图像文字方向检测
+@author: lywen
+"""

angle/predict.py

@@ -0,0 +1,49 @@
+# -*- coding: utf-8 -*-
+# from keras.models import load_model
+import numpy as np
+from PIL import Image
+from keras.applications.vgg16 import preprocess_input, VGG16
+from keras.layers import Dense
+from keras.models import Model
+##编译模型，以较小的学习参数进行训练
+from keras.optimizers import SGD
+
+
+def load():
+    vgg = VGG16(weights=None, input_shape=(224, 224, 3))
+    ##修改输出层 3个输出
+    x = vgg.layers[-2].output
+    predictions_class = Dense(4, activation='softmax', name='predictions_class')(x)
+    prediction = [predictions_class]
+    model = Model(inputs=vgg.input, outputs=prediction)
+    sgd = SGD(lr=0.00001, momentum=0.9)
+    model.compile(optimizer=sgd, loss='categorical_crossentropy', metrics=['accuracy'])
+    model.load_weights('/Users/xiaofeng/Code/Github/dataset/CHINESE_OCR/angle/modelAngle.h5')
+    return model
+
+
+##加载模型
+model = None
+
+
+def predict(path=None, img=None):
+    global model
+    if model is None:
+        model = load()
+    """
+    图片文字方向预测
+    """
+    ROTATE = [0, 90, 180, 270]
+    if path is not None:
+        im = Image.open(path).convert('RGB')
+    elif img is not None:
+        im = Image.fromarray(img).convert('RGB')
+    w, h = im.size
+    xmin, ymin, xmax, ymax = int(0.1 * w), int(0.1 * h), w - int(0.1 * w), h - int(0.1 * h)
+    im = im.crop((xmin, ymin, xmax, ymax))  ##剪切图片边缘，清楚边缘噪声
+    im = im.resize((224, 224))
+    img = np.array(im)
+    img = preprocess_input(img.astype(np.float32))
+    pred = model.predict(np.array([img]))
+    index = np.argmax(pred, axis=1)[0]
+    return ROTATE[index]