使用PaddleOCR识别文字（二）代码实现篇

刺骨的言语ヽ痛彻心扉 2022-12-26 11:20 174阅读 0赞

vue页面axios上传图片到后台，python处理上传及识别，识别记录插入数据库。  
代码还没拆分，后续优化时候再拆。  
server.py

```python
    #coding=utf8
    import web,sys,os  #引入web.py的包
    class MyApplication(web.application):
        def run(self, port=8080, *middleware):
            func = self.wsgifunc(*middleware)
            return web.httpserver.runsimple(func, ('0.0.0.0', port))
     
    #定义web应用的路由规则
    urls = (
        '/discern/word', 'word'  # 解析图片
    )
     
    class word:
        def POST(self):
            import time,random,json
            from PIL import Image
            from paddleocr import PaddleOCR, draw_ocr
            from db_config import db_config
    
            year = time.strftime('%Y',time.localtime(time.time()))
            month = time.strftime('%m',time.localtime(time.time()))
            aiBaseDir = os.path.realpath('./../../server/upload/ai/'+ year+'/'+ month)   # 上传的原始图片基础路径 dir/ai/2020/12
            aiResultDir = os.path.realpath('./../../server/upload/airesult/'+ year+'/'+ month)   # 识别之后的结果图片基础路径 dir/airesult/2020/12
            if not os.path.exists(aiBaseDir):
                os.makedirs( aiBaseDir )
            if not os.path.exists(aiResultDir):
                os.makedirs( aiResultDir )
    
            inp = web.input(myfile={},type="")  # post的值
    
            discernType = inp.type  #识别的类型
    
            #上传图片
            filename = str(time.time()) + str(random.randint(11111,999999)) + inp['myfile'].filename[inp['myfile'].filename.rfind('.'):]  # 文件名
    
            dbImgUrl = '/' + year + '/' + month + '/' + filename  # 数据库中存的图片路径 /2020/12/a.jpg
    
            baseFullPath = aiBaseDir +'/'+ filename  #上传的文件全路径
            resultFullPath = aiResultDir +'/'+ filename  #识别之后的文件全路径
            fout = open(baseFullPath,'wb')
            fout.write(inp.myfile.file.read())
            fout.write(inp.myfile.value)
            fout.close()
    
            ocr = PaddleOCR(use_angle_cls=True, langs="ch")
            result = ocr.ocr(baseFullPath, cls=True)
    
            # for line in result:
            #     print(line)
            # print(result)
    
            image = Image.open(baseFullPath).convert('RGB')
            boxes = [line[0] for line in result]
            txts = [line[1][0] for line in result]
            scores = [line[1][1] for line in result]
            imShow = draw_ocr(image, boxes, txts, scores, font_path=os.path.realpath('./../ai/simfang.ttf'))
            imShow = Image.fromarray(imShow)
            imShow.save(resultFullPath)
            web.header('Content-Type','text/json; charset=utf-8', unique=True) #防止中文乱码
            web.header("Access-Control-Allow-Origin", "*", unique=True) #允许跨域
            db = web.database(dbn=db_config['dbn'],host=db_config['host'], user=db_config['user'], pw=db_config['pwd'], db=db_config['database'])
            content = []
            for i in result:
                a = {}
                a['title'] = str(i[1][0])
                a['percent'] = str(i[1][1])
                a['position'] = i[0]
                if a['title']:
                    content.append(a)
            insertId = db.insert("T_AI_LOG",IMG = dbImgUrl,IMG_NAME=inp.myfile.filename,TYPE = discernType,CONTENT= json.dumps(content),DATE=time.strftime('%Y-%m-%d %H:%M:%S',time.localtime(time.time())))
            if insertId:
                return True
            else:
                return False
    if __name__ == "__main__":
        app = MyApplication(urls, globals())
        app.run(port=6789)
    ```

安装nohup

yum install coreutils

后台启动脚本

nohup python3 server.py >/dev/null &