PaddleOCR加载chinese_ocr_db_crnn_modile模型进行中英文混合预测（Http服务）实践

比眉伴天荒 2022-11-10 05:33 162阅读 0赞

##### 1. 环境搭建 #####

参考：《[PaddleOCR加载chinese\_ocr\_db\_crnn\_server模型进行中英文混合预测（命令行）实践][PaddleOCR_chinese_ocr_db_crnn_server]》

##### 2. 服务端部署 #####

hub serving start -m chinese_ocr_db_crnn_mobile -p 8866

##### 3. 客户端访问 #####

# coding: utf8
    import requests
    import json
    import cv2
    import base64
    
    def cv2_to_base64(image):
        data = cv2.imencode('.jpg', image)[1]
        return base64.b64encode(data.tostring()).decode('utf8')
    
    # 发送HTTP请求
    data = { 'images':[cv2_to_base64(cv2.imread("/PATH/TO/IMAGE"))]}
    headers = { "Content-type": "application/json"}
    url = "http://127.0.0.1:8866/predict/chinese_ocr_db_crnn_mobile"
    r = requests.post(url=url, headers=headers, data=json.dumps(data))
    
    # 打印预测结果
    print(r.json()["results"])

##### 4. 将图像识别服务部署到Flask上 #####

先参考《[Flask开发实践][Flask]》或《[Flask入门][Flask 1]》  
在此基础上进行了以下修改：

文件routes.py增加的内容

import base64
    import json
    import os
    import cv2
    import requests
    from werkzeug.utils import secure_filename
    
    
    # OCR测试
    @webapp.route('/ocr', methods=['POST', 'GET'])
    def ocr():
        basepath = os.path.dirname(__file__)  # 当前文件所在路径
        print(request.method)
        if request.method == 'POST':
            f = request.files['file']
            upload_path = os.path.join(basepath, 'static/uploads', secure_filename(f.filename))  # 注意：没有的文件夹要先创建
            f.save(upload_path)
            return redirect(url_for('ocr_result', pic_name=f.filename))
        else:
            return render_template('ocr.html')
    
    
    # OCR测试结果
    @webapp.route('/ocr/<pic_name>', methods=['POST', 'GET'])
    def ocr_result(pic_name):
        basepath = os.path.dirname(__file__)  # 当前文件所在路径
        print(request.method)
        if pic_name != "":
            upload_path = os.path.join(basepath, 'static/uploads', secure_filename(pic_name))
            data = { 'images': [cv2_to_base64(cv2.imread(upload_path))]}
            headers = { "Content-type": "application/json"}
            url = "http://127.0.0.1:8866/predict/chinese_ocr_db_crnn_mobile"
            # 图像识别服务： hub.exe serving start -m chinese_ocr_db_crnn_mobile -p 8866
            r = requests.post(url=url, headers=headers, data=json.dumps(data))
            # 打印预测结果
            results = r.json()["results"][0]["data"]
    
            return render_template('ocr.html', results=results, file=upload_path.replace(basepath, ''))
        else:
            return render_template('ocr.html')
    
    
    def cv2_to_base64(image):
        data = cv2.imencode('.jpg', image)[1]
        return base64.b64encode(data.tostring()).decode('utf8')

新建模板文件ocr.html

<!DOCTYPE html>
    <html >
    <head>
        <meta charset="UTF-8">
            <title>图像识别测试页</title>
    </head>
    <body>
        <h1>图像识别测试页</h1>
        <form action="/ocr" enctype="multipart/form-data" method="post">
            <input type="file" name="file">
            <input type="submit" value="提交">
        </form>
        <h1>识别结果</h1>
        <img src="{ { file }}">
        {% for result in results %}
            <table>
                <tr valign="top">
                    <td>[置信度] {
       { (result.confidence*100) | round(2) }}%, [文字内容] {
       { result.text }}</td>
                </tr>
            </table>
        {% endfor %}
        {
       { result }}
    </body>
    </html>

##### 5. 运行效果 #####

先上传图片  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZtMDUxNw_size_16_color_FFFFFF_t_70_pic_center]

提交后显示图片内容和识别结果  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZtMDUxNw_size_16_color_FFFFFF_t_70_pic_center 1]

【参考文档】  
《[PaddleHub一键OCR中文识别][PaddleHub_OCR]》

[PaddleOCR_chinese_ocr_db_crnn_server]: https://blog.csdn.net/fm0517/article/details/114867135
[Flask]: https://blog.csdn.net/fm0517/article/details/113884899
[Flask 1]: https://blog.csdn.net/fm0517/article/details/113867029
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZtMDUxNw_size_16_color_FFFFFF_t_70_pic_center]: /images/20221023/494b6fe5ba0640d18e96b8927ac686c1.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2ZtMDUxNw_size_16_color_FFFFFF_t_70_pic_center 1]: /images/20221023/f5b0eba96b0f41208dd4c54e04a4539a.png
[PaddleHub_OCR]: https://aistudio.baidu.com/aistudio/projectdetail/507159