diff --git a/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/README.md b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..2076a10cc32f3c99562a89ff8d933585003b5169
--- /dev/null
+++ b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/README.md
@@ -0,0 +1,125 @@
+## 概述
+
+chinese_ocr_db_rcnn Module用于识别图片当中的汉字。其基于[chinese_text_detection_db Module](https://www.paddlepaddle.org.cn/hubdetail?name=chinese_text_detection_db&en_category=TextRecognition)检测得到的文本框，继续识别文本框中的中文文字。识别文字算法采用CRNN（Convolutional Recurrent Neural Network）即卷积递归神经网络。其是DCNN和RNN的组合，专门用于识别图像中的序列式对象。与CTC loss配合使用，进行文字识别，可以直接从文本词级或行级的标注中学习，不需要详细的字符级的标注。该Module支持直接预测。
+
+
+
+
+
+
+更多详情参考[An end-to-end trainable neural network for image-based sequence recognition and its application to scene text recognition](https://arxiv.org/pdf/1507.05717.pdf)
+
+## 命令行预测
+
+```shell
+$ hub run chinese_ocr_db_rcnn --input_path "/PATH/TO/IMAGE"
+```
+
+## API
+
+```python
+def recognize_text(images=[],
+ paths=[],
+ use_gpu=False,
+ output_dir='ocr_result',
+ visualization=False,
+ box_thresh=0.5,
+ text_thresh=0.5)
+```
+
+预测API，检测输入图片中的所有中文文本的位置。
+
+**参数**
+
+* paths (list\[str\]): 图片的路径；
+* images (list\[numpy.ndarray\]): 图片数据，ndarray.shape 为 \[H, W, C\]，BGR格式；
+* use\_gpu (bool): 是否使用 GPU；**若使用GPU，请先设置CUDA_VISIBLE_DEVICES环境变量**
+* box\_thresh (float): 检测文本框置信度的阈值；
+* text\_thresh (float): 识别中文文本置信度的阈值；
+* visualization (bool): 是否将识别结果保存为图片文件；
+* output\_dir (str): 图片的保存路径，默认设为 detection\_result；
+
+**返回**
+
+* res (list\[dict\]): 识别结果的列表，列表中每一个元素为 dict，各字段为：
+ * data (list\[dict\]): 识别文本结果，列表中每一个元素为 dict，各字段为：
+ * text(str): 识别得到的文本
+ * confidence(float): 识别文本结果置信度
+ * text_box_position(numpy.ndarray): 文本框在原图中的像素坐标，4*2的矩阵，依次表示文本框左下、右下、右上、左上顶点的坐标
+ 如果无识别结果则data为\[\]
+ * save_path (str, optional): 识别结果的保存路径，如不保存图片则save_path为''
+
+### 代码示例
+
+```python
+import paddlehub as hub
+import cv2
+
+ocr = hub.Module(name="chinese_ocr_db_rcnn")
+result = ocr.recognize_text(images=[cv2.imread('/PATH/TO/IMAGE')])
+
+# or
+# result = ocr.recognize_text(paths=['/PATH/TO/IMAGE'])
+```
+
+* 样例结果示例
+
+
+
+
+
+## 服务部署
+
+PaddleHub Serving 可以部署一个目标检测的在线服务。
+
+### 第一步：启动PaddleHub Serving
+
+运行启动命令：
+```shell
+$ hub serving start -m chinese_ocr_db_rcnn
+```
+
+这样就完成了一个目标检测的服务化API的部署，默认端口号为8866。
+
+**NOTE:** 如使用GPU预测，则需要在启动服务之前，请设置CUDA\_VISIBLE\_DEVICES环境变量，否则不用设置。
+
+### 第二步：发送预测请求
+
+配置好服务端，以下数行代码即可实现发送预测请求，获取预测结果
+
+```python
+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_rcnn"
+r = requests.post(url=url, headers=headers, data=json.dumps(data))
+
+# 打印预测结果
+print(r.json()["results"])
+```
+
+## 查看代码
+
+https://github.com/PaddlePaddle/PaddleOCR
+
+### 依赖
+
+paddlepaddle >= 1.7.2
+
+paddlehub >= 1.6.0
+
+
+## 更新历史
+
+* 1.0.0
+
+ 初始发布
diff --git a/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/__init__.py b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/assets/ppocr_keys_v1.txt b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/assets/ppocr_keys_v1.txt
new file mode 100644
index 0000000000000000000000000000000000000000..b75af2130342e619dbb9f3f87dc8b74aa27b4a76
--- /dev/null
+++ b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/assets/ppocr_keys_v1.txt
@@ -0,0 +1,6623 @@
+'
+疗
+绚
+诚
+娇
+溜
+题
+贿
+者
+廖
+更
+纳
+加
+奉
+公
+一
+就
+汴
+计
+与
+路
+房
+原
+妇
+2
+0
+8
+-
+7
+其
+>
+:
+]
+,
+，
+骑
+刈
+全
+消
+昏
+傈
+安
+久
+钟
+嗅
+不
+影
+处
+驽
+蜿
+资
+关
+椤
+地
+瘸
+专
+问
+忖
+票
+嫉
+炎
+韵
+要
+月
+田
+节
+陂
+鄙
+捌
+备
+拳
+伺
+眼
+网
+盎
+大
+傍
+心
+东
+愉
+汇
+蹿
+科
+每
+业
+里
+航
+晏
+字
+平
+录
+先
+1
+3
+彤
+鲶
+产
+稍
+督
+腴
+有
+象
+岳
+注
+绍
+在
+泺
+文
+定
+核
+名
+水
+过
+理
+让
+偷
+率
+等
+这
+发
+”
+为
+含
+肥
+酉
+相
+鄱
+七
+编
+猥
+锛
+日
+镀
+蒂
+掰
+倒
+辆
+栾
+栗
+综
+涩
+州
+雌
+滑
+馀
+了
+机
+块
+司
+宰
+甙
+兴
+矽
+抚
+保
+用
+沧
+秩
+如
+收
+息
+滥
+页
+疑
+埠
+!
+！
+姥
+异
+橹
+钇
+向
+下
+跄
+的
+椴
+沫
+国
+绥
+獠
+报
+开
+民
+蜇
+何
+分
+凇
+长
+讥
+藏
+掏
+施
+羽
+中
+讲
+派
+嘟
+人
+提
+浼
+间
+世
+而
+古
+多
+倪
+唇
+饯
+控
+庚
+首
+赛
+蜓
+味
+断
+制
+觉
+技
+替
+艰
+溢
+潮
+夕
+钺
+外
+摘
+枋
+动
+双
+单
+啮
+户
+枇
+确
+锦
+曜
+杜
+或
+能
+效
+霜
+盒
+然
+侗
+电
+晁
+放
+步
+鹃
+新
+杖
+蜂
+吒
+濂
+瞬
+评
+总
+隍
+对
+独
+合
+也
+是
+府
+青
+天
+诲
+墙
+组
+滴
+级
+邀
+帘
+示
+已
+时
+骸
+仄
+泅
+和
+遨
+店
+雇
+疫
+持
+巍
+踮
+境
+只
+亨
+目
+鉴
+崤
+闲
+体
+泄
+杂
+作
+般
+轰
+化
+解
+迂
+诿
+蛭
+璀
+腾
+告
+版
+服
+省
+师
+小
+规
+程
+线
+海
+办
+引
+二
+桧
+牌
+砺
+洄
+裴
+修
+图
+痫
+胡
+许
+犊
+事
+郛
+基
+柴
+呼
+食
+研
+奶
+律
+蛋
+因
+葆
+察
+戏
+褒
+戒
+再
+李
+骁
+工
+貂
+油
+鹅
+章
+啄
+休
+场
+给
+睡
+纷
+豆
+器
+捎
+说
+敏
+学
+会
+浒
+设
+诊
+格
+廓
+查
+来
+霓
+室
+溆
+￠
+诡
+寥
+焕
+舜
+柒
+狐
+回
+戟
+砾
+厄
+实
+翩
+尿
+五
+入
+径
+惭
+喹
+股
+宇
+篝
+|
+;
+美
+期
+云
+九
+祺
+扮
+靠
+锝
+槌
+系
+企
+酰
+阊
+暂
+蚕
+忻
+豁
+本
+羹
+执
+条
+钦
+H
+獒
+限
+进
+季
+楦
+于
+芘
+玖
+铋
+茯
+未
+答
+粘
+括
+样
+精
+欠
+矢
+甥
+帷
+嵩
+扣
+令
+仔
+风
+皈
+行
+支
+部
+蓉
+刮
+站
+蜡
+救
+钊
+汗
+松
+嫌
+成
+可
+.
+鹤
+院
+从
+交
+政
+怕
+活
+调
+球
+局
+验
+髌
+第
+韫
+谗
+串
+到
+圆
+年
+米
+/
+*
+友
+忿
+检
+区
+看
+自
+敢
+刃
+个
+兹
+弄
+流
+留
+同
+没
+齿
+星
+聆
+轼
+湖
+什
+三
+建
+蛔
+儿
+椋
+汕
+震
+颧
+鲤
+跟
+力
+情
+璺
+铨
+陪
+务
+指
+族
+训
+滦
+鄣
+濮
+扒
+商
+箱
+十
+召
+慷
+辗
+所
+莞
+管
+护
+臭
+横
+硒
+嗓
+接
+侦
+六
+露
+党
+馋
+驾
+剖
+高
+侬
+妪
+幂
+猗
+绺
+骐
+央
+酐
+孝
+筝
+课
+徇
+缰
+门
+男
+西
+项
+句
+谙
+瞒
+秃
+篇
+教
+碲
+罚
+声
+呐
+景
+前
+富
+嘴
+鳌
+稀
+免
+朋
+啬
+睐
+去
+赈
+鱼
+住
+肩
+愕
+速
+旁
+波
+厅
+健
+茼
+厥
+鲟
+谅
+投
+攸
+炔
+数
+方
+击
+呋
+谈
+绩
+别
+愫
+僚
+躬
+鹧
+胪
+炳
+招
+喇
+膨
+泵
+蹦
+毛
+结
+5
+4
+谱
+识
+陕
+粽
+婚
+拟
+构
+且
+搜
+任
+潘
+比
+郢
+妨
+醪
+陀
+桔
+碘
+扎
+选
+哈
+骷
+楷
+亿
+明
+缆
+脯
+监
+睫
+逻
+婵
+共
+赴
+淝
+凡
+惦
+及
+达
+揖
+谩
+澹
+减
+焰
+蛹
+番
+祁
+柏
+员
+禄
+怡
+峤
+龙
+白
+叽
+生
+闯
+起
+细
+装
+谕
+竟
+聚
+钙
+上
+导
+渊
+按
+艾
+辘
+挡
+耒
+盹
+饪
+臀
+记
+邮
+蕙
+受
+各
+医
+搂
+普
+滇
+朗
+茸
+带
+翻
+酚
+(
+光
+堤
+墟
+蔷
+万
+幻
+〓
+瑙
+辈
+昧
+盏
+亘
+蛀
+吉
+铰
+请
+子
+假
+闻
+税
+井
+诩
+哨
+嫂
+好
+面
+琐
+校
+馊
+鬣
+缂
+营
+访
+炖
+占
+农
+缀
+否
+经
+钚
+棵
+趟
+张
+亟
+吏
+茶
+谨
+捻
+论
+迸
+堂
+玉
+信
+吧
+瞠
+乡
+姬
+寺
+咬
+溏
+苄
+皿
+意
+赉
+宝
+尔
+钰
+艺
+特
+唳
+踉
+都
+荣
+倚
+登
+荐
+丧
+奇
+涵
+批
+炭
+近
+符
+傩
+感
+道
+着
+菊
+虹
+仲
+众
+懈
+濯
+颞
+眺
+南
+释
+北
+缝
+标
+既
+茗
+整
+撼
+迤
+贲
+挎
+耱
+拒
+某
+妍
+卫
+哇
+英
+矶
+藩
+治
+他
+元
+领
+膜
+遮
+穗
+蛾
+飞
+荒
+棺
+劫
+么
+市
+火
+温
+拈
+棚
+洼
+转
+果
+奕
+卸
+迪
+伸
+泳
+斗
+邡
+侄
+涨
+屯
+萋
+胭
+氡
+崮
+枞
+惧
+冒
+彩
+斜
+手
+豚
+随
+旭
+淑
+妞
+形
+菌
+吲
+沱
+争
+驯
+歹
+挟
+兆
+柱
+传
+至
+包
+内
+响
+临
+红
+功
+弩
+衡
+寂
+禁
+老
+棍
+耆
+渍
+织
+害
+氵
+渑
+布
+载
+靥
+嗬
+虽
+苹
+咨
+娄
+库
+雉
+榜
+帜
+嘲
+套
+瑚
+亲
+簸
+欧
+边
+6
+腿
+旮
+抛
+吹
+瞳
+得
+镓
+梗
+厨
+继
+漾
+愣
+憨
+士
+策
+窑
+抑
+躯
+襟
+脏
+参
+贸
+言
+干
+绸
+鳄
+穷
+藜
+音
+折
+详
+)
+举
+悍
+甸
+癌
+黎
+谴
+死
+罩
+迁
+寒
+驷
+袖
+媒
+蒋
+掘
+模
+纠
+恣
+观
+祖
+蛆
+碍
+位
+稿
+主
+澧
+跌
+筏
+京
+锏
+帝
+贴
+证
+糠
+才
+黄
+鲸
+略
+炯
+饱
+四
+出
+园
+犀
+牧
+容
+汉
+杆
+浈
+汰
+瑷
+造
+虫
+瘩
+怪
+驴
+济
+应
+花
+沣
+谔
+夙
+旅
+价
+矿
+以
+考
+s
+u
+呦
+晒
+巡
+茅
+准
+肟
+瓴
+詹
+仟
+褂
+译
+桌
+混
+宁
+怦
+郑
+抿
+些
+余
+鄂
+饴
+攒
+珑
+群
+阖
+岔
+琨
+藓
+预
+环
+洮
+岌
+宀
+杲
+瀵
+最
+常
+囡
+周
+踊
+女
+鼓
+袭
+喉
+简
+范
+薯
+遐
+疏
+粱
+黜
+禧
+法
+箔
+斤
+遥
+汝
+奥
+直
+贞
+撑
+置
+绱
+集
+她
+馅
+逗
+钧
+橱
+魉
+[
+恙
+躁
+唤
+9
+旺
+膘
+待
+脾
+惫
+购
+吗
+依
+盲
+度
+瘿
+蠖
+俾
+之
+镗
+拇
+鲵
+厝
+簧
+续
+款
+展
+啃
+表
+剔
+品
+钻
+腭
+损
+清
+锶
+统
+涌
+寸
+滨
+贪
+链
+吠
+冈
+伎
+迥
+咏
+吁
+览
+防
+迅
+失
+汾
+阔
+逵
+绀
+蔑
+列
+川
+凭
+努
+熨
+揪
+利
+俱
+绉
+抢
+鸨
+我
+即
+责
+膦
+易
+毓
+鹊
+刹
+玷
+岿
+空
+嘞
+绊
+排
+术
+估
+锷
+违
+们
+苟
+铜
+播
+肘
+件
+烫
+审
+鲂
+广
+像
+铌
+惰
+铟
+巳
+胍
+鲍
+康
+憧
+色
+恢
+想
+拷
+尤
+疳
+知
+S
+Y
+F
+D
+A
+峄
+裕
+帮
+握
+搔
+氐
+氘
+难
+墒
+沮
+雨
+叁
+缥
+悴
+藐
+湫
+娟
+苑
+稠
+颛
+簇
+后
+阕
+闭
+蕤
+缚
+怎
+佞
+码
+嘤
+蔡
+痊
+舱
+螯
+帕
+赫
+昵
+升
+烬
+岫
+、
+疵
+蜻
+髁
+蕨
+隶
+烛
+械
+丑
+盂
+梁
+强
+鲛
+由
+拘
+揉
+劭
+龟
+撤
+钩
+呕
+孛
+费
+妻
+漂
+求
+阑
+崖
+秤
+甘
+通
+深
+补
+赃
+坎
+床
+啪
+承
+吼
+量
+暇
+钼
+烨
+阂
+擎
+脱
+逮
+称
+P
+神
+属
+矗
+华
+届
+狍
+葑
+汹
+育
+患
+窒
+蛰
+佼
+静
+槎
+运
+鳗
+庆
+逝
+曼
+疱
+克
+代
+官
+此
+麸
+耧
+蚌
+晟
+例
+础
+榛
+副
+测
+唰
+缢
+迹
+灬
+霁
+身
+岁
+赭
+扛
+又
+菡
+乜
+雾
+板
+读
+陷
+徉
+贯
+郁
+虑
+变
+钓
+菜
+圾
+现
+琢
+式
+乐
+维
+渔
+浜
+左
+吾
+脑
+钡
+警
+T
+啵
+拴
+偌
+漱
+湿
+硕
+止
+骼
+魄
+积
+燥
+联
+踢
+玛
+则
+窿
+见
+振
+畿
+送
+班
+钽
+您
+赵
+刨
+印
+讨
+踝
+籍
+谡
+舌
+崧
+汽
+蔽
+沪
+酥
+绒
+怖
+财
+帖
+肱
+私
+莎
+勋
+羔
+霸
+励
+哼
+帐
+将
+帅
+渠
+纪
+婴
+娩
+岭
+厘
+滕
+吻
+伤
+坝
+冠
+戊
+隆
+瘁
+介
+涧
+物
+黍
+并
+姗
+奢
+蹑
+掣
+垸
+锴
+命
+箍
+捉
+病
+辖
+琰
+眭
+迩
+艘
+绌
+繁
+寅
+若
+毋
+思
+诉
+类
+诈
+燮
+轲
+酮
+狂
+重
+反
+职
+筱
+县
+委
+磕
+绣
+奖
+晋
+濉
+志
+徽
+肠
+呈
+獐
+坻
+口
+片
+碰
+几
+村
+柿
+劳
+料
+获
+亩
+惕
+晕
+厌
+号
+罢
+池
+正
+鏖
+煨
+家
+棕
+复
+尝
+懋
+蜥
+锅
+岛
+扰
+队
+坠
+瘾
+钬
+@
+卧
+疣
+镇
+譬
+冰
+彷
+频
+黯
+据
+垄
+采
+八
+缪
+瘫
+型
+熹
+砰
+楠
+襁
+箐
+但
+嘶
+绳
+啤
+拍
+盥
+穆
+傲
+洗
+盯
+塘
+怔
+筛
+丿
+台
+恒
+喂
+葛
+永
+￥
+烟
+酒
+桦
+书
+砂
+蚝
+缉
+态
+瀚
+袄
+圳
+轻
+蛛
+超
+榧
+遛
+姒
+奘
+铮
+右
+荽
+望
+偻
+卡
+丶
+氰
+附
+做
+革
+索
+戚
+坨
+桷
+唁
+垅
+榻
+岐
+偎
+坛
+莨
+山
+殊
+微
+骇
+陈
+爨
+推
+嗝
+驹
+澡
+藁
+呤
+卤
+嘻
+糅
+逛
+侵
+郓
+酌
+德
+摇
+※
+鬃
+被
+慨
+殡
+羸
+昌
+泡
+戛
+鞋
+河
+宪
+沿
+玲
+鲨
+翅
+哽
+源
+铅
+语
+照
+邯
+址
+荃
+佬
+顺
+鸳
+町
+霭
+睾
+瓢
+夸
+椁
+晓
+酿
+痈
+咔
+侏
+券
+噎
+湍
+签
+嚷
+离
+午
+尚
+社
+锤
+背
+孟
+使
+浪
+缦
+潍
+鞅
+军
+姹
+驶
+笑
+鳟
+鲁
+》
+孽
+钜
+绿
+洱
+礴
+焯
+椰
+颖
+囔
+乌
+孔
+巴
+互
+性
+椽
+哞
+聘
+昨
+早
+暮
+胶
+炀
+隧
+低
+彗
+昝
+铁
+呓
+氽
+藉
+喔
+癖
+瑗
+姨
+权
+胱
+韦
+堑
+蜜
+酋
+楝
+砝
+毁
+靓
+歙
+锲
+究
+屋
+喳
+骨
+辨
+碑
+武
+鸠
+宫
+辜
+烊
+适
+坡
+殃
+培
+佩
+供
+走
+蜈
+迟
+翼
+况
+姣
+凛
+浔
+吃
+飘
+债
+犟
+金
+促
+苛
+崇
+坂
+莳
+畔
+绂
+兵
+蠕
+斋
+根
+砍
+亢
+欢
+恬
+崔
+剁
+餐
+榫
+快
+扶
+‖
+濒
+缠
+鳜
+当
+彭
+驭
+浦
+篮
+昀
+锆
+秸
+钳
+弋
+娣
+瞑
+夷
+龛
+苫
+拱
+致
+%
+嵊
+障
+隐
+弑
+初
+娓
+抉
+汩
+累
+蓖
+"
+唬
+助
+苓
+昙
+押
+毙
+破
+城
+郧
+逢
+嚏
+獭
+瞻
+溱
+婿
+赊
+跨
+恼
+璧
+萃
+姻
+貉
+灵
+炉
+密
+氛
+陶
+砸
+谬
+衔
+点
+琛
+沛
+枳
+层
+岱
+诺
+脍
+榈
+埂
+征
+冷
+裁
+打
+蹴
+素
+瘘
+逞
+蛐
+聊
+激
+腱
+萘
+踵
+飒
+蓟
+吆
+取
+咙
+簋
+涓
+矩
+曝
+挺
+揣
+座
+你
+史
+舵
+焱
+尘
+苏
+笈
+脚
+溉
+榨
+诵
+樊
+邓
+焊
+义
+庶
+儋
+蟋
+蒲
+赦
+呷
+杞
+诠
+豪
+还
+试
+颓
+茉
+太
+除
+紫
+逃
+痴
+草
+充
+鳕
+珉
+祗
+墨
+渭
+烩
+蘸
+慕
+璇
+镶
+穴
+嵘
+恶
+骂
+险
+绋
+幕
+碉
+肺
+戳
+刘
+潞
+秣
+纾
+潜
+銮
+洛
+须
+罘
+销
+瘪
+汞
+兮
+屉
+r
+林
+厕
+质
+探
+划
+狸
+殚
+善
+煊
+烹
+〒
+锈
+逯
+宸
+辍
+泱
+柚
+袍
+远
+蹋
+嶙
+绝
+峥
+娥
+缍
+雀
+徵
+认
+镱
+谷
+=
+贩
+勉
+撩
+鄯
+斐
+洋
+非
+祚
+泾
+诒
+饿
+撬
+威
+晷
+搭
+芍
+锥
+笺
+蓦
+候
+琊
+档
+礁
+沼
+卵
+荠
+忑
+朝
+凹
+瑞
+头
+仪
+弧
+孵
+畏
+铆
+突
+衲
+车
+浩
+气
+茂
+悖
+厢
+枕
+酝
+戴
+湾
+邹
+飚
+攘
+锂
+写
+宵
+翁
+岷
+无
+喜
+丈
+挑
+嗟
+绛
+殉
+议
+槽
+具
+醇
+淞
+笃
+郴
+阅
+饼
+底
+壕
+砚
+弈
+询
+缕
+庹
+翟
+零
+筷
+暨
+舟
+闺
+甯
+撞
+麂
+茌
+蔼
+很
+珲
+捕
+棠
+角
+阉
+媛
+娲
+诽
+剿
+尉
+爵
+睬
+韩
+诰
+匣
+危
+糍
+镯
+立
+浏
+阳
+少
+盆
+舔
+擘
+匪
+申
+尬
+铣
+旯
+抖
+赘
+瓯
+居
+ˇ
+哮
+游
+锭
+茏
+歌
+坏
+甚
+秒
+舞
+沙
+仗
+劲
+潺
+阿
+燧
+郭
+嗖
+霏
+忠
+材
+奂
+耐
+跺
+砀
+输
+岖
+媳
+氟
+极
+摆
+灿
+今
+扔
+腻
+枝
+奎
+药
+熄
+吨
+话
+q
+额
+慑
+嘌
+协
+喀
+壳
+埭
+视
+著
+於
+愧
+陲
+翌
+峁
+颅
+佛
+腹
+聋
+侯
+咎
+叟
+秀
+颇
+存
+较
+罪
+哄
+岗
+扫
+栏
+钾
+羌
+己
+璨
+枭
+霉
+煌
+涸
+衿
+键
+镝
+益
+岢
+奏
+连
+夯
+睿
+冥
+均
+糖
+狞
+蹊
+稻
+爸
+刿
+胥
+煜
+丽
+肿
+璃
+掸
+跚
+灾
+垂
+樾
+濑
+乎
+莲
+窄
+犹
+撮
+战
+馄
+软
+络
+显
+鸢
+胸
+宾
+妲
+恕
+埔
+蝌
+份
+遇
+巧
+瞟
+粒
+恰
+剥
+桡
+博
+讯
+凯
+堇
+阶
+滤
+卖
+斌
+骚
+彬
+兑
+磺
+樱
+舷
+两
+娱
+福
+仃
+差
+找
+桁
+÷
+净
+把
+阴
+污
+戬
+雷
+碓
+蕲
+楚
+罡
+焖
+抽
+妫
+咒
+仑
+闱
+尽
+邑
+菁
+爱
+贷
+沥
+鞑
+牡
+嗉
+崴
+骤
+塌
+嗦
+订
+拮
+滓
+捡
+锻
+次
+坪
+杩
+臃
+箬
+融
+珂
+鹗
+宗
+枚
+降
+鸬
+妯
+阄
+堰
+盐
+毅
+必
+杨
+崃
+俺
+甬
+状
+莘
+货
+耸
+菱
+腼
+铸
+唏
+痤
+孚
+澳
+懒
+溅
+翘
+疙
+杷
+淼
+缙
+骰
+喊
+悉
+砻
+坷
+艇
+赁
+界
+谤
+纣
+宴
+晃
+茹
+归
+饭
+梢
+铡
+街
+抄
+肼
+鬟
+苯
+颂
+撷
+戈
+炒
+咆
+茭
+瘙
+负
+仰
+客
+琉
+铢
+封
+卑
+珥
+椿
+镧
+窨
+鬲
+寿
+御
+袤
+铃
+萎
+砖
+餮
+脒
+裳
+肪
+孕
+嫣
+馗
+嵇
+恳
+氯
+江
+石
+褶
+冢
+祸
+阻
+狈
+羞
+银
+靳
+透
+咳
+叼
+敷
+芷
+啥
+它
+瓤
+兰
+痘
+懊
+逑
+肌
+往
+捺
+坊
+甩
+呻
+〃
+沦
+忘
+膻
+祟
+菅
+剧
+崆
+智
+坯
+臧
+霍
+墅
+攻
+眯
+倘
+拢
+骠
+铐
+庭
+岙
+瓠
+′
+缺
+泥
+迢
+捶
+?
+？
+郏
+喙
+掷
+沌
+纯
+秘
+种
+听
+绘
+固
+螨
+团
+香
+盗
+妒
+埚
+蓝
+拖
+旱
+荞
+铀
+血
+遏
+汲
+辰
+叩
+拽
+幅
+硬
+惶
+桀
+漠
+措
+泼
+唑
+齐
+肾
+念
+酱
+虚
+屁
+耶
+旗
+砦
+闵
+婉
+馆
+拭
+绅
+韧
+忏
+窝
+醋
+葺
+顾
+辞
+倜
+堆
+辋
+逆
+玟
+贱
+疾
+董
+惘
+倌
+锕
+淘
+嘀
+莽
+俭
+笏
+绑
+鲷
+杈
+择
+蟀
+粥
+嗯
+驰
+逾
+案
+谪
+褓
+胫
+哩
+昕
+颚
+鲢
+绠
+躺
+鹄
+崂
+儒
+俨
+丝
+尕
+泌
+啊
+萸
+彰
+幺
+吟
+骄
+苣
+弦
+脊
+瑰
+〈
+诛
+镁
+析
+闪
+剪
+侧
+哟
+框
+螃
+守
+嬗
+燕
+狭
+铈
+缮
+概
+迳
+痧
+鲲
+俯
+售
+笼
+痣
+扉
+挖
+满
+咋
+援
+邱
+扇
+歪
+便
+玑
+绦
+峡
+蛇
+叨
+〖
+泽
+胃
+斓
+喋
+怂
+坟
+猪
+该
+蚬
+炕
+弥
+赞
+棣
+晔
+娠
+挲
+狡
+创
+疖
+铕
+镭
+稷
+挫
+弭
+啾
+翔
+粉
+履
+苘
+哦
+楼
+秕
+铂
+土
+锣
+瘟
+挣
+栉
+习
+享
+桢
+袅
+磨
+桂
+谦
+延
+坚
+蔚
+噗
+署
+谟
+猬
+钎
+恐
+嬉
+雒
+倦
+衅
+亏
+璩
+睹
+刻
+殿
+王
+算
+雕
+麻
+丘
+柯
+骆
+丸
+塍
+谚
+添
+鲈
+垓
+桎
+蚯
+芥
+予
+飕
+镦
+谌
+窗
+醚
+菀
+亮
+搪
+莺
+蒿
+羁
+足
+J
+真
+轶
+悬
+衷
+靛
+翊
+掩
+哒
+炅
+掐
+冼
+妮
+l
+谐
+稚
+荆
+擒
+犯
+陵
+虏
+浓
+崽
+刍
+陌
+傻
+孜
+千
+靖
+演
+矜
+钕
+煽
+杰
+酗
+渗
+伞
+栋
+俗
+泫
+戍
+罕
+沾
+疽
+灏
+煦
+芬
+磴
+叱
+阱
+榉
+湃
+蜀
+叉
+醒
+彪
+租
+郡
+篷
+屎
+良
+垢
+隗
+弱
+陨
+峪
+砷
+掴
+颁
+胎
+雯
+绵
+贬
+沐
+撵
+隘
+篙
+暖
+曹
+陡
+栓
+填
+臼
+彦
+瓶
+琪
+潼
+哪
+鸡
+摩
+啦
+俟
+锋
+域
+耻
+蔫
+疯
+纹
+撇
+毒
+绶
+痛
+酯
+忍
+爪
+赳
+歆
+嘹
+辕
+烈
+册
+朴
+钱
+吮
+毯
+癜
+娃
+谀
+邵
+厮
+炽
+璞
+邃
+丐
+追
+词
+瓒
+忆
+轧
+芫
+谯
+喷
+弟
+半
+冕
+裙
+掖
+墉
+绮
+寝
+苔
+势
+顷
+褥
+切
+衮
+君
+佳
+嫒
+蚩
+霞
+佚
+洙
+逊
+镖
+暹
+唛
+&
+殒
+顶
+碗
+獗
+轭
+铺
+蛊
+废
+恹
+汨
+崩
+珍
+那
+杵
+曲
+纺
+夏
+薰
+傀
+闳
+淬
+姘
+舀
+拧
+卷
+楂
+恍
+讪
+厩
+寮
+篪
+赓
+乘
+灭
+盅
+鞣
+沟
+慎
+挂
+饺
+鼾
+杳
+树
+缨
+丛
+絮
+娌
+臻
+嗳
+篡
+侩
+述
+衰
+矛
+圈
+蚜
+匕
+筹
+匿
+濞
+晨
+叶
+骋
+郝
+挚
+蚴
+滞
+增
+侍
+描
+瓣
+吖
+嫦
+蟒
+匾
+圣
+赌
+毡
+癞
+恺
+百
+曳
+需
+篓
+肮
+庖
+帏
+卿
+驿
+遗
+蹬
+鬓
+骡
+歉
+芎
+胳
+屐
+禽
+烦
+晌
+寄
+媾
+狄
+翡
+苒
+船
+廉
+终
+痞
+殇
+々
+畦
+饶
+改
+拆
+悻
+萄
+￡
+瓿
+乃
+訾
+桅
+匮
+溧
+拥
+纱
+铍
+骗
+蕃
+龋
+缬
+父
+佐
+疚
+栎
+醍
+掳
+蓄
+x
+惆
+颜
+鲆
+榆
+〔
+猎
+敌
+暴
+谥
+鲫
+贾
+罗
+玻
+缄
+扦
+芪
+癣
+落
+徒
+臾
+恿
+猩
+托
+邴
+肄
+牵
+春
+陛
+耀
+刊
+拓
+蓓
+邳
+堕
+寇
+枉
+淌
+啡
+湄
+兽
+酷
+萼
+碚
+濠
+萤
+夹
+旬
+戮
+梭
+琥
+椭
+昔
+勺
+蜊
+绐
+晚
+孺
+僵
+宣
+摄
+冽
+旨
+萌
+忙
+蚤
+眉
+噼
+蟑
+付
+契
+瓜
+悼
+颡
+壁
+曾
+窕
+颢
+澎
+仿
+俑
+浑
+嵌
+浣
+乍
+碌
+褪
+乱
+蔟
+隙
+玩
+剐
+葫
+箫
+纲
+围
+伐
+决
+伙
+漩
+瑟
+刑
+肓
+镳
+缓
+蹭
+氨
+皓
+典
+畲
+坍
+铑
+檐
+塑
+洞
+倬
+储
+胴
+淳
+戾
+吐
+灼
+惺
+妙
+毕
+珐
+缈
+虱
+盖
+羰
+鸿
+磅
+谓
+髅
+娴
+苴
+唷
+蚣
+霹
+抨
+贤
+唠
+犬
+誓
+逍
+庠
+逼
+麓
+籼
+釉
+呜
+碧
+秧
+氩
+摔
+霄
+穸
+纨
+辟
+妈
+映
+完
+牛
+缴
+嗷
+炊
+恩
+荔
+茆
+掉
+紊
+慌
+莓
+羟
+阙
+萁
+磐
+另
+蕹
+辱
+鳐
+湮
+吡
+吩
+唐
+睦
+垠
+舒
+圜
+冗
+瞿
+溺
+芾
+囱
+匠
+僳
+汐
+菩
+饬
+漓
+黑
+霰
+浸
+濡
+窥
+毂
+蒡
+兢
+驻
+鹉
+芮
+诙
+迫
+雳
+厂
+忐
+臆
+猴
+鸣
+蚪
+栈
+箕
+羡
+渐
+莆
+捍
+眈
+哓
+趴
+蹼
+埕
+嚣
+骛
+宏
+淄
+斑
+噜
+严
+瑛
+垃
+椎
+诱
+压
+庾
+绞
+焘
+廿
+抡
+迄
+棘
+夫
+纬
+锹
+眨
+瞌
+侠
+脐
+竞
+瀑
+孳
+骧
+遁
+姜
+颦
+荪
+滚
+萦
+伪
+逸
+粳
+爬
+锁
+矣
+役
+趣
+洒
+颔
+诏
+逐
+奸
+甭
+惠
+攀
+蹄
+泛
+尼
+拼
+阮
+鹰
+亚
+颈
+惑
+勒
+〉
+际
+肛
+爷
+刚
+钨
+丰
+养
+冶
+鲽
+辉
+蔻
+画
+覆
+皴
+妊
+麦
+返
+醉
+皂
+擀
+〗
+酶
+凑
+粹
+悟
+诀
+硖
+港
+卜
+z
+杀
+涕
+±
+舍
+铠
+抵
+弛
+段
+敝
+镐
+奠
+拂
+轴
+跛
+袱
+e
+t
+沉
+菇
+俎
+薪
+峦
+秭
+蟹
+历
+盟
+菠
+寡
+液
+肢
+喻
+染
+裱
+悱
+抱
+氙
+赤
+捅
+猛
+跑
+氮
+谣
+仁
+尺
+辊
+窍
+烙
+衍
+架
+擦
+倏
+璐
+瑁
+币
+楞
+胖
+夔
+趸
+邛
+惴
+饕
+虔
+蝎
+§
+哉
+贝
+宽
+辫
+炮
+扩
+饲
+籽
+魏
+菟
+锰
+伍
+猝
+末
+琳
+哚
+蛎
+邂
+呀
+姿
+鄞
+却
+歧
+仙
+恸
+椐
+森
+牒
+寤
+袒
+婆
+虢
+雅
+钉
+朵
+贼
+欲
+苞
+寰
+故
+龚
+坭
+嘘
+咫
+礼
+硷
+兀
+睢
+汶
+’
+铲
+烧
+绕
+诃
+浃
+钿
+哺
+柜
+讼
+颊
+璁
+腔
+洽
+咐
+脲
+簌
+筠
+镣
+玮
+鞠
+谁
+兼
+姆
+挥
+梯
+蝴
+谘
+漕
+刷
+躏
+宦
+弼
+b
+垌
+劈
+麟
+莉
+揭
+笙
+渎
+仕
+嗤
+仓
+配
+怏
+抬
+错
+泯
+镊
+孰
+猿
+邪
+仍
+秋
+鼬
+壹
+歇
+吵
+炼
+<
+尧
+射
+柬
+廷
+胧
+霾
+凳
+隋
+肚
+浮
+梦
+祥
+株
+堵
+退
+L
+鹫
+跎
+凶
+毽
+荟
+炫
+栩
+玳
+甜
+沂
+鹿
+顽
+伯
+爹
+赔
+蛴
+徐
+匡
+欣
+狰
+缸
+雹
+蟆
+疤
+默
+沤
+啜
+痂
+衣
+禅
+w
+i
+h
+辽
+葳
+黝
+钗
+停
+沽
+棒
+馨
+颌
+肉
+吴
+硫
+悯
+劾
+娈
+马
+啧
+吊
+悌
+镑
+峭
+帆
+瀣
+涉
+咸
+疸
+滋
+泣
+翦
+拙
+癸
+钥
+蜒
++
+尾
+庄
+凝
+泉
+婢
+渴
+谊
+乞
+陆
+锉
+糊
+鸦
+淮
+I
+B
+N
+晦
+弗
+乔
+庥
+葡
+尻
+席
+橡
+傣
+渣
+拿
+惩
+麋
+斛
+缃
+矮
+蛏
+岘
+鸽
+姐
+膏
+催
+奔
+镒
+喱
+蠡
+摧
+钯
+胤
+柠
+拐
+璋
+鸥
+卢
+荡
+倾
+^
+_
+珀
+逄
+萧
+塾
+掇
+贮
+笆
+聂
+圃
+冲
+嵬
+M
+滔
+笕
+值
+炙
+偶
+蜱
+搐
+梆
+汪
+蔬
+腑
+鸯
+蹇
+敞
+绯
+仨
+祯
+谆
+梧
+糗
+鑫
+啸
+豺
+囹
+猾
+巢
+柄
+瀛
+筑
+踌
+沭
+暗
+苁
+鱿
+蹉
+脂
+蘖
+牢
+热
+木
+吸
+溃
+宠
+序
+泞
+偿
+拜
+檩
+厚
+朐
+毗
+螳
+吞
+媚
+朽
+担
+蝗
+橘
+畴
+祈
+糟
+盱
+隼
+郜
+惜
+珠
+裨
+铵
+焙
+琚
+唯
+咚
+噪
+骊
+丫
+滢
+勤
+棉
+呸
+咣
+淀
+隔
+蕾
+窈
+饨
+挨
+煅
+短
+匙
+粕
+镜
+赣
+撕
+墩
+酬
+馁
+豌
+颐
+抗
+酣
+氓
+佑
+搁
+哭
+递
+耷
+涡
+桃
+贻
+碣
+截
+瘦
+昭
+镌
+蔓
+氚
+甲
+猕
+蕴
+蓬
+散
+拾
+纛
+狼
+猷
+铎
+埋
+旖
+矾
+讳
+囊
+糜
+迈
+粟
+蚂
+紧
+鲳
+瘢
+栽
+稼
+羊
+锄
+斟
+睁
+桥
+瓮
+蹙
+祉
+醺
+鼻
+昱
+剃
+跳
+篱
+跷
+蒜
+翎
+宅
+晖
+嗑
+壑
+峻
+癫
+屏
+狠
+陋
+袜
+途
+憎
+祀
+莹
+滟
+佶
+溥
+臣
+约
+盛
+峰
+磁
+慵
+婪
+拦
+莅
+朕
+鹦
+粲
+裤
+哎
+疡
+嫖
+琵
+窟
+堪
+谛
+嘉
+儡
+鳝
+斩
+郾
+驸
+酊
+妄
+胜
+贺
+徙
+傅
+噌
+钢
+栅
+庇
+恋
+匝
+巯
+邈
+尸
+锚
+粗
+佟
+蛟
+薹
+纵
+蚊
+郅
+绢
+锐
+苗
+俞
+篆
+淆
+膀
+鲜
+煎
+诶
+秽
+寻
+涮
+刺
+怀
+噶
+巨
+褰
+魅
+灶
+灌
+桉
+藕
+谜
+舸
+薄
+搀
+恽
+借
+牯
+痉
+渥
+愿
+亓
+耘
+杠
+柩
+锔
+蚶
+钣
+珈
+喘
+蹒
+幽
+赐
+稗
+晤
+莱
+泔
+扯
+肯
+菪
+裆
+腩
+豉
+疆
+骜
+腐
+倭
+珏
+唔
+粮
+亡
+润
+慰
+伽
+橄
+玄
+誉
+醐
+胆
+龊
+粼
+塬
+陇
+彼
+削
+嗣
+绾
+芽
+妗
+垭
+瘴
+爽
+薏
+寨
+龈
+泠
+弹
+赢
+漪
+猫
+嘧
+涂
+恤
+圭
+茧
+烽
+屑
+痕
+巾
+赖
+荸
+凰
+腮
+畈
+亵
+蹲
+偃
+苇
+澜
+艮
+换
+骺
+烘
+苕
+梓
+颉
+肇
+哗
+悄
+氤
+涠
+葬
+屠
+鹭
+植
+竺
+佯
+诣
+鲇
+瘀
+鲅
+邦
+移
+滁
+冯
+耕
+癔
+戌
+茬
+沁
+巩
+悠
+湘
+洪
+痹
+锟
+循
+谋
+腕
+鳃
+钠
+捞
+焉
+迎
+碱
+伫
+急
+榷
+奈
+邝
+卯
+辄
+皲
+卟
+醛
+畹
+忧
+稳
+雄
+昼
+缩
+阈
+睑
+扌
+耗
+曦
+涅
+捏
+瞧
+邕
+淖
+漉
+铝
+耦
+禹
+湛
+喽
+莼
+琅
+诸
+苎
+纂
+硅
+始
+嗨
+傥
+燃
+臂
+赅
+嘈
+呆
+贵
+屹
+壮
+肋
+亍
+蚀
+卅
+豹
+腆
+邬
+迭
+浊
+}
+童
+螂
+捐
+圩
+勐
+触
+寞
+汊
+壤
+荫
+膺
+渌
+芳
+懿
+遴
+螈
+泰
+蓼
+蛤
+茜
+舅
+枫
+朔
+膝
+眙
+避
+梅
+判
+鹜
+璜
+牍
+缅
+垫
+藻
+黔
+侥
+惚
+懂
+踩
+腰
+腈
+札
+丞
+唾
+慈
+顿
+摹
+荻
+琬
+~
+斧
+沈
+滂
+胁
+胀
+幄
+莜
+Z
+匀
+鄄
+掌
+绰
+茎
+焚
+赋
+萱
+谑
+汁
+铒
+瞎
+夺
+蜗
+野
+娆
+冀
+弯
+篁
+懵
+灞
+隽
+芡
+脘
+俐
+辩
+芯
+掺
+喏
+膈
+蝈
+觐
+悚
+踹
+蔗
+熠
+鼠
+呵
+抓
+橼
+峨
+畜
+缔
+禾
+崭
+弃
+熊
+摒
+凸
+拗
+穹
+蒙
+抒
+祛
+劝
+闫
+扳
+阵
+醌
+踪
+喵
+侣
+搬
+仅
+荧
+赎
+蝾
+琦
+买
+婧
+瞄
+寓
+皎
+冻
+赝
+箩
+莫
+瞰
+郊
+笫
+姝
+筒
+枪
+遣
+煸
+袋
+舆
+痱
+涛
+母
+〇
+启
+践
+耙
+绲
+盘
+遂
+昊
+搞
+槿
+诬
+纰
+泓
+惨
+檬
+亻
+越
+C
+o
+憩
+熵
+祷
+钒
+暧
+塔
+阗
+胰
+咄
+娶
+魔
+琶
+钞
+邻
+扬
+杉
+殴
+咽
+弓
+〆
+髻
+】
+吭
+揽
+霆
+拄
+殖
+脆
+彻
+岩
+芝
+勃
+辣
+剌
+钝
+嘎
+甄
+佘
+皖
+伦
+授
+徕
+憔
+挪
+皇
+庞
+稔
+芜
+踏
+溴
+兖
+卒
+擢
+饥
+鳞
+煲
+‰
+账
+颗
+叻
+斯
+捧
+鳍
+琮
+讹
+蛙
+纽
+谭
+酸
+兔
+莒
+睇
+伟
+觑
+羲
+嗜
+宜
+褐
+旎
+辛
+卦
+诘
+筋
+鎏
+溪
+挛
+熔
+阜
+晰
+鳅
+丢
+奚
+灸
+呱
+献
+陉
+黛
+鸪
+甾
+萨
+疮
+拯
+洲
+疹
+辑
+叙
+恻
+谒
+允
+柔
+烂
+氏
+逅
+漆
+拎
+惋
+扈
+湟
+纭
+啕
+掬
+擞
+哥
+忽
+涤
+鸵
+靡
+郗
+瓷
+扁
+廊
+怨
+雏
+钮
+敦
+E
+懦
+憋
+汀
+拚
+啉
+腌
+岸
+f
+痼
+瞅
+尊
+咀
+眩
+飙
+忌
+仝
+迦
+熬
+毫
+胯
+篑
+茄
+腺
+凄
+舛
+碴
+锵
+诧
+羯
+後
+漏
+汤
+宓
+仞
+蚁
+壶
+谰
+皑
+铄
+棰
+罔
+辅
+晶
+苦
+牟
+闽
+\
+烃
+饮
+聿
+丙
+蛳
+朱
+煤
+涔
+鳖
+犁
+罐
+荼
+砒
+淦
+妤
+黏
+戎
+孑
+婕
+瑾
+戢
+钵
+枣
+捋
+砥
+衩
+狙
+桠
+稣
+阎
+肃
+梏
+诫
+孪
+昶
+婊
+衫
+嗔
+侃
+塞
+蜃
+樵
+峒
+貌
+屿
+欺
+缫
+阐
+栖
+诟
+珞
+荭
+吝
+萍
+嗽
+恂
+啻
+蜴
+磬
+峋
+俸
+豫
+谎
+徊
+镍
+韬
+魇
+晴
+U
+囟
+猜
+蛮
+坐
+囿
+伴
+亭
+肝
+佗
+蝠
+妃
+胞
+滩
+榴
+氖
+垩
+苋
+砣
+扪
+馏
+姓
+轩
+厉
+夥
+侈
+禀
+垒
+岑
+赏
+钛
+辐
+痔
+披
+纸
+碳
+“
+坞
+蠓
+挤
+荥
+沅
+悔
+铧
+帼
+蒌
+蝇
+a
+p
+y
+n
+g
+哀
+浆
+瑶
+凿
+桶
+馈
+皮
+奴
+苜
+佤
+伶
+晗
+铱
+炬
+优
+弊
+氢
+恃
+甫
+攥
+端
+锌
+灰
+稹
+炝
+曙
+邋
+亥
+眶
+碾
+拉
+萝
+绔
+捷
+浍
+腋
+姑
+菖
+凌
+涞
+麽
+锢
+桨
+潢
+绎
+镰
+殆
+锑
+渝
+铬
+困
+绽
+觎
+匈
+糙
+暑
+裹
+鸟
+盔
+肽
+迷
+綦
+『
+亳
+佝
+俘
+钴
+觇
+骥
+仆
+疝
+跪
+婶
+郯
+瀹
+唉
+脖
+踞
+针
+晾
+忒
+扼
+瞩
+叛
+椒
+疟
+嗡
+邗
+肆
+跆
+玫
+忡
+捣
+咧
+唆
+艄
+蘑
+潦
+笛
+阚
+沸
+泻
+掊
+菽
+贫
+斥
+髂
+孢
+镂
+赂
+麝
+鸾
+屡
+衬
+苷
+恪
+叠
+希
+粤
+爻
+喝
+茫
+惬
+郸
+绻
+庸
+撅
+碟
+宄
+妹
+膛
+叮
+饵
+崛
+嗲
+椅
+冤
+搅
+咕
+敛
+尹
+垦
+闷
+蝉
+霎
+勰
+败
+蓑
+泸
+肤
+鹌
+幌
+焦
+浠
+鞍
+刁
+舰
+乙
+竿
+裔
+。
+茵
+函
+伊
+兄
+丨
+娜
+匍
+謇
+莪
+宥
+似
+蝽
+翳
+酪
+翠
+粑
+薇
+祢
+骏
+赠
+叫
+Q
+噤
+噻
+竖
+芗
+莠
+潭
+俊
+羿
+耜
+O
+郫
+趁
+嗪
+囚
+蹶
+芒
+洁
+笋
+鹑
+敲
+硝
+啶
+堡
+渲
+揩
+』
+携
+宿
+遒
+颍
+扭
+棱
+割
+萜
+蔸
+葵
+琴
+捂
+饰
+衙
+耿
+掠
+募
+岂
+窖
+涟
+蔺
+瘤
+柞
+瞪
+怜
+匹
+距
+楔
+炜
+哆
+秦
+缎
+幼
+茁
+绪
+痨
+恨
+楸
+娅
+瓦
+桩
+雪
+嬴
+伏
+榔
+妥
+铿
+拌
+眠
+雍
+缇
+‘
+卓
+搓
+哌
+觞
+噩
+屈
+哧
+髓
+咦
+巅
+娑
+侑
+淫
+膳
+祝
+勾
+姊
+莴
+胄
+疃
+薛
+蜷
+胛
+巷
+芙
+芋
+熙
+闰
+勿
+窃
+狱
+剩
+钏
+幢
+陟
+铛
+慧
+靴
+耍
+k
+浙
+浇
+飨
+惟
+绗
+祜
+澈
+啼
+咪
+磷
+摞
+诅
+郦
+抹
+跃
+壬
+吕
+肖
+琏
+颤
+尴
+剡
+抠
+凋
+赚
+泊
+津
+宕
+殷
+倔
+氲
+漫
+邺
+涎
+怠
+$
+垮
+荬
+遵
+俏
+叹
+噢
+饽
+蜘
+孙
+筵
+疼
+鞭
+羧
+牦
+箭
+潴
+c
+眸
+祭
+髯
+啖
+坳
+愁
+芩
+驮
+倡
+巽
+穰
+沃
+胚
+怒
+凤
+槛
+剂
+趵
+嫁
+v
+邢
+灯
+鄢
+桐
+睽
+檗
+锯
+槟
+婷
+嵋
+圻
+诗
+蕈
+颠
+遭
+痢
+芸
+怯
+馥
+竭
+锗
+徜
+恭
+遍
+籁
+剑
+嘱
+苡
+龄
+僧
+桑
+潸
+弘
+澶
+楹
+悲
+讫
+愤
+腥
+悸
+谍
+椹
+呢
+桓
+葭
+攫
+阀
+翰
+躲
+敖
+柑
+郎
+笨
+橇
+呃
+魁
+燎
+脓
+葩
+磋
+垛
+玺
+狮
+沓
+砜
+蕊
+锺
+罹
+蕉
+翱
+虐
+闾
+巫
+旦
+茱
+嬷
+枯
+鹏
+贡
+芹
+汛
+矫
+绁
+拣
+禺
+佃
+讣
+舫
+惯
+乳
+趋
+疲
+挽
+岚
+虾
+衾
+蠹
+蹂
+飓
+氦
+铖
+孩
+稞
+瑜
+壅
+掀
+勘
+妓
+畅
+髋
+W
+庐
+牲
+蓿
+榕
+练
+垣
+唱
+邸
+菲
+昆
+婺
+穿
+绡
+麒
+蚱
+掂
+愚
+泷
+涪
+漳
+妩
+娉
+榄
+讷
+觅
+旧
+藤
+煮
+呛
+柳
+腓
+叭
+庵
+烷
+阡
+罂
+蜕
+擂
+猖
+咿
+媲
+脉
+【
+沏
+貅
+黠
+熏
+哲
+烁
+坦
+酵
+兜
+×
+潇
+撒
+剽
+珩
+圹
+乾
+摸
+樟
+帽
+嗒
+襄
+魂
+轿
+憬
+锡
+〕
+喃
+皆
+咖
+隅
+脸
+残
+泮
+袂
+鹂
+珊
+囤
+捆
+咤
+误
+徨
+闹
+淙
+芊
+淋
+怆
+囗
+拨
+梳
+渤
+R
+G
+绨
+蚓
+婀
+幡
+狩
+麾
+谢
+唢
+裸
+旌
+伉
+纶
+裂
+驳
+砼
+咛
+澄
+樨
+蹈
+宙
+澍
+倍
+貔
+操
+勇
+蟠
+摈
+砧
+虬
+够
+缁
+悦
+藿
+撸
+艹
+摁
+淹
+豇
+虎
+榭
+ˉ
+吱
+d
+°
+喧
+荀
+踱
+侮
+奋
+偕
+饷
+犍
+惮
+坑
+璎
+徘
+宛
+妆
+袈
+倩
+窦
+昂
+荏
+乖
+K
+怅
+撰
+鳙
+牙
+袁
+酞
+X
+痿
+琼
+闸
+雁
+趾
+荚
+虻
+涝
+《
+杏
+韭
+偈
+烤
+绫
+鞘
+卉
+症
+遢
+蓥
+诋
+杭
+荨
+匆
+竣
+簪
+辙
+敕
+虞
+丹
+缭
+咩
+黟
+m
+淤
+瑕
+咂
+铉
+硼
+茨
+嶂
+痒
+畸
+敬
+涿
+粪
+窘
+熟
+叔
+嫔
+盾
+忱
+裘
+憾
+梵
+赡
+珙
+咯
+娘
+庙
+溯
+胺
+葱
+痪
+摊
+荷
+卞
+乒
+髦
+寐
+铭
+坩
+胗
+枷
+爆
+溟
+嚼
+羚
+砬
+轨
+惊
+挠
+罄
+竽
+菏
+氧
+浅
+楣
+盼
+枢
+炸
+阆
+杯
+谏
+噬
+淇
+渺
+俪
+秆
+墓
+泪
+跻
+砌
+痰
+垡
+渡
+耽
+釜
+讶
+鳎
+煞
+呗
+韶
+舶
+绷
+鹳
+缜
+旷
+铊
+皱
+龌
+檀
+霖
+奄
+槐
+艳
+蝶
+旋
+哝
+赶
+骞
+蚧
+腊
+盈
+丁
+`
+蜚
+矸
+蝙
+睨
+嚓
+僻
+鬼
+醴
+夜
+彝
+磊
+笔
+拔
+栀
+糕
+厦
+邰
+纫
+逭
+纤
+眦
+膊
+馍
+躇
+烯
+蘼
+冬
+诤
+暄
+骶
+哑
+瘠
+」
+臊
+丕
+愈
+咱
+螺
+擅
+跋
+搏
+硪
+谄
+笠
+淡
+嘿
+骅
+谧
+鼎
+皋
+姚
+歼
+蠢
+驼
+耳
+胬
+挝
+涯
+狗
+蒽
+孓
+犷
+凉
+芦
+箴
+铤
+孤
+嘛
+坤
+V
+茴
+朦
+挞
+尖
+橙
+诞
+搴
+碇
+洵
+浚
+帚
+蜍
+漯
+柘
+嚎
+讽
+芭
+荤
+咻
+祠
+秉
+跖
+埃
+吓
+糯
+眷
+馒
+惹
+娼
+鲑
+嫩
+讴
+轮
+瞥
+靶
+褚
+乏
+缤
+宋
+帧
+删
+驱
+碎
+扑
+俩
+俄
+偏
+涣
+竹
+噱
+皙
+佰
+渚
+唧
+斡
+#
+镉
+刀
+崎
+筐
+佣
+夭
+贰
+肴
+峙
+哔
+艿
+匐
+牺
+镛
+缘
+仡
+嫡
+劣
+枸
+堀
+梨
+簿
+鸭
+蒸
+亦
+稽
+浴
+{
+衢
+束
+槲
+j
+阁
+揍
+疥
+棋
+潋
+聪
+窜
+乓
+睛
+插
+冉
+阪
+苍
+搽
+「
+蟾
+螟
+幸
+仇
+樽
+撂
+慢
+跤
+幔
+俚
+淅
+覃
+觊
+溶
+妖
+帛
+侨
+曰
+妾
+泗
+·
+：
+瀘
+風
+Ë
+（
+）
+∶
+紅
+紗
+瑭
+雲
+頭
+鶏
+財
+許
+•
+¥
+樂
+焗
+麗
+—
+；
+滙
+東
+榮
+繪
+興
+…
+門
+業
+π
+楊
+國
+顧
+é
+盤
+寳
+Λ
+龍
+鳳
+島
+誌
+緣
+結
+銭
+萬
+勝
+祎
+璟
+優
+歡
+臨
+時
+購
+＝
+★
+藍
+昇
+鐵
+觀
+勅
+農
+聲
+畫
+兿
+術
+發
+劉
+記
+專
+耑
+園
+書
+壴
+種
+Ο
+●
+褀
+號
+銀
+匯
+敟
+锘
+葉
+橪
+廣
+進
+蒄
+鑽
+阝
+祙
+貢
+鍋
+豊
+夬
+喆
+團
+閣
+開
+燁
+賓
+館
+酡
+沔
+順
+＋
+硚
+劵
+饸
+陽
+車
+湓
+復
+萊
+氣
+軒
+華
+堃
+迮
+纟
+戶
+馬
+學
+裡
+電
+嶽
+獨
+マ
+シ
+サ
+ジ
+燘
+袪
+環
+❤
+臺
+灣
+専
+賣
+孖
+聖
+攝
+線
+▪
+α
+傢
+俬
+夢
+達
+莊
+喬
+貝
+薩
+劍
+羅
+壓
+棛
+饦
+尃
+璈
+囍
+醫
+Ｇ
+Ｉ
+Ａ
+＃
+Ｎ
+鷄
+髙
+嬰
+啓
+約
+隹
+潔
+賴
+藝
+～
+寶
+籣
+麺
+
+嶺
+√
+義
+網
+峩
+長
+∧
+魚
+機
+構
+②
+鳯
+偉
+Ｌ
+Ｂ
+㙟
+畵
+鴿
+＇
+詩
+溝
+嚞
+屌
+藔
+佧
+玥
+蘭
+織
+１
+３
+９
+０
+７
+點
+砭
+鴨
+鋪
+銘
+廳
+弍
+‧
+創
+湯
+坶
+℃
+卩
+骝
+＆
+烜
+荘
+當
+潤
+扞
+係
+懷
+碶
+钅
+蚨
+讠
+☆
+叢
+爲
+埗
+涫
+塗
+→
+楽
+現
+鯨
+愛
+瑪
+鈺
+忄
+悶
+藥
+飾
+樓
+視
+孬
+ㆍ
+燚
+苪
+師
+①
+丼
+锽
+│
+韓
+標
+è
+兒
+閏
+匋
+張
+漢
+Ü
+髪
+會
+閑
+檔
+習
+裝
+の
+峯
+菘
+輝
+И
+雞
+釣
+億
+浐
+Ｋ
+Ｏ
+Ｒ
+８
+Ｈ
+Ｅ
+Ｐ
+Ｔ
+Ｗ
+Ｄ
+Ｓ
+Ｃ
+Ｍ
+Ｆ
+姌
+饹
+»
+晞
+廰
+ä
+嵯
+鷹
+負
+飲
+絲
+冚
+楗
+澤
+綫
+區
+❋
+←
+質
+靑
+揚
+③
+滬
+統
+産
+協
+﹑
+乸
+畐
+經
+運
+際
+洺
+岽
+為
+粵
+諾
+崋
+豐
+碁
+ɔ
+Ｖ
+２
+６
+齋
+誠
+訂
+´
+勑
+雙
+陳
+無
+í
+泩
+媄
+夌
+刂
+ｉ
+ｃ
+ｔ
+ｏ
+ｒ
+ａ
+嘢
+耄
+燴
+暃
+壽
+媽
+靈
+抻
+體
+唻
+É
+冮
+甹
+鎮
+錦
+ʌ
+蜛
+蠄
+尓
+駕
+戀
+飬
+逹
+倫
+貴
+極
+Я
+Й
+寬
+磚
+嶪
+郎
+職
+｜
+間
+ｎ
+ｄ
+剎
+伈
+課
+飛
+橋
+瘊
+№
+譜
+骓
+圗
+滘
+縣
+粿
+咅
+養
+濤
+彳
+®
+％
+Ⅱ
+啰
+㴪
+見
+矞
+薬
+糁
+邨
+鲮
+顔
+罱
+З
+選
+話
+贏
+氪
+俵
+競
+瑩
+繡
+枱
+β
+綉
+á
+獅
+爾
+™
+麵
+戋
+淩
+徳
+個
+劇
+場
+務
+簡
+寵
+ｈ
+實
+膠
+轱
+圖
+築
+嘣
+樹
+㸃
+營
+耵
+孫
+饃
+鄺
+飯
+麯
+遠
+輸
+坫
+孃
+乚
+閃
+鏢
+㎡
+題
+廠
+關
+↑
+爺
+將
+軍
+連
+篦
+覌
+參
+箸
+－
+窠
+棽
+寕
+夀
+爰
+歐
+呙
+閥
+頡
+熱
+雎
+垟
+裟
+凬
+勁
+帑
+馕
+夆
+疌
+枼
+馮
+貨
+蒤
+樸
+彧
+旸
+靜
+龢
+暢
+㐱
+鳥
+珺
+鏡
+灡
+爭
+堷
+廚
+Ó
+騰
+診
+┅
+蘇
+褔
+凱
+頂
+豕
+亞
+帥
+嘬
+⊥
+仺
+桖
+複
+饣
+絡
+穂
+顏
+棟
+納
+▏
+濟
+親
+設
+計
+攵
+埌
+烺
+ò
+頤
+燦
+蓮
+撻
+節
+講
+濱
+濃
+娽
+洳
+朿
+燈
+鈴
+護
+膚
+铔
+過
+補
+Ｚ
+Ｕ
+５
+４
+坋
+闿
+䖝
+餘
+缐
+铞
+貿
+铪
+桼
+趙
+鍊
+［
+㐂
+垚
+菓
+揸
+捲
+鐘
+滏
+𣇉
+爍
+輪
+燜
+鴻
+鮮
+動
+鹞
+鷗
+丄
+慶
+鉌
+翥
+飮
+腸
+⇋
+漁
+覺
+來
+熘
+昴
+翏
+鲱
+圧
+鄉
+萭
+頔
+爐
+嫚
+г
+貭
+類
+聯
+幛
+輕
+訓
+鑒
+夋
+锨
+芃
+珣
+䝉
+扙
+嵐
+銷
+處
+ㄱ
+語
+誘
+苝
+歸
+儀
+燒
+楿
+內
+粢
+葒
+奧
+麥
+礻
+滿
+蠔
+穵
+瞭
+態
+鱬
+榞
+硂
+鄭
+黃
+煙
+祐
+奓
+逺
+＊
+瑄
+獲
+聞
+薦
+讀
+這
+樣
+決
+問
+啟
+們
+執
+説
+轉
+單
+隨
+唘
+帶
+倉
+庫
+還
+贈
+尙
+皺
+■
+餅
+產
+○
+∈
+報
+狀
+楓
+賠
+琯
+嗮
+禮
+｀
+傳
+＞
+≤
+嗞
+Φ
+≥
+換
+咭
+∣
+↓
+曬
+ε
+応
+寫
+″
+終
+様
+純
+費
+療
+聨
+凍
+壐
+郵
+ü
+黒
+∫
+製
+塊
+調
+軽
+確
+撃
+級
+馴
+Ⅲ
+涇
+繹
+數
+碼
+證
+狒
+処
+劑
+＜
+晧
+賀
+衆
+］
+櫥
+兩
+陰
+絶
+對
+鯉
+憶
+◎
+ｐ
+ｅ
+Ｙ
+蕒
+煖
+頓
+測
+試
+鼽
+僑
+碩
+妝
+帯
+≈
+鐡
+舖
+權
+喫
+倆
+ˋ
+該
+悅
+ā
+俫
+．
+ｆ
+ｓ
+ｂ
+ｍ
+ｋ
+ｇ
+ｕ
+ｊ
+貼
+淨
+濕
+針
+適
+備
+ｌ
+／
+給
+謢
+強
+觸
+衛
+與
+⊙
+＄
+緯
+變
+⑴
+⑵
+⑶
+㎏
+殺
+∩
+幚
+─
+價
+▲
+離
+ú
+ó
+飄
+烏
+関
+閟
+﹝
+﹞
+邏
+輯
+鍵
+驗
+訣
+導
+歷
+屆
+層
+▼
+儱
+錄
+熳
+ē
+艦
+吋
+錶
+辧
+飼
+顯
+④
+禦
+販
+気
+対
+枰
+閩
+紀
+幹
+瞓
+貊
+淚
+△
+眞
+墊
+Ω
+獻
+褲
+縫
+緑
+亜
+鉅
+餠
+｛
+｝
+◆
+蘆
+薈
+█
+◇
+溫
+彈
+晳
+粧
+犸
+穩
+訊
+崬
+凖
+熥
+П
+舊
+條
+紋
+圍
+Ⅳ
+筆
+尷
+難
+雜
+錯
+綁
+識
+頰
+鎖
+艶
+□
+殁
+殼
+⑧
+├
+▕
+鵬
+ǐ
+ō
+ǒ
+糝
+綱
+▎
+μ
+盜
+饅
+醬
+籤
+蓋
+釀
+鹽
+據
+à
+ɡ
+辦
+◥
+彐
+┌
+婦
+獸
+鲩
+伱
+ī
+蒟
+蒻
+齊
+袆
+腦
+寧
+凈
+妳
+煥
+詢
+偽
+謹
+啫
+鯽
+騷
+鱸
+損
+傷
+鎻
+髮
+買
+冏
+儥
+両
+﹢
+∞
+載
+喰
+ｚ
+羙
+悵
+燙
+曉
+員
+組
+徹
+艷
+痠
+鋼
+鼙
+縮
+細
+嚒
+爯
+≠
+維
+＂
+鱻
+壇
+厍
+帰
+浥
+犇
+薡
+軎
+²
+應
+醜
+刪
+緻
+鶴
+賜
+噁
+軌
+尨
+镔
+鷺
+槗
+彌
+葚
+濛
+請
+溇
+緹
+賢
+訪
+獴
+瑅
+資
+縤
+陣
+蕟
+栢
+韻
+祼
+恁
+伢
+謝
+劃
+涑
+總
+衖
+踺
+砋
+凉
+籃
+駿
+苼
+瘋
+昽
+紡
+驊
+腎
+﹗
+響
+杋
+剛
+嚴
+禪
+歓
+槍
+傘
+檸
+檫
+炣
+勢
+鏜
+鎢
+銑
+尐
+減
+奪
+惡
+θ
+僮
+婭
+臘
+ū
+ì
+殻
+鉄
+∑
+蛲
+焼
+緖
+續
+紹
+懮
diff --git a/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/assets/simfang.ttf b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/assets/simfang.ttf
new file mode 100644
index 0000000000000000000000000000000000000000..2b59eae4195d1cdbea375503c0cc34d5631cb0f9
Binary files /dev/null and b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/assets/simfang.ttf differ
diff --git a/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/character.py b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/character.py
new file mode 100644
index 0000000000000000000000000000000000000000..8e5f10211ba441a7dd9b4948413b79c8721eab07
--- /dev/null
+++ b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/character.py
@@ -0,0 +1,168 @@
+# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import numpy as np
+import string
+
+
+class CharacterOps(object):
+ """ Convert between text-label and text-index """
+
+ def __init__(self, config):
+ self.character_type = config['character_type']
+ self.loss_type = config['loss_type']
+ if self.character_type == "en":
+ self.character_str = "0123456789abcdefghijklmnopqrstuvwxyz"
+ dict_character = list(self.character_str)
+ elif self.character_type == "ch":
+ character_dict_path = config['character_dict_path']
+ self.character_str = ""
+ with open(character_dict_path, "rb") as fin:
+ lines = fin.readlines()
+ for line in lines:
+ line = line.decode('utf-8').strip("\n")
+ self.character_str += line
+ dict_character = list(self.character_str)
+ elif self.character_type == "en_sensitive":
+ # same with ASTER setting (use 94 char).
+ self.character_str = string.printable[:-6]
+ dict_character = list(self.character_str)
+ else:
+ self.character_str = None
+ assert self.character_str is not None, \
+ "Nonsupport type of the character: {}".format(self.character_str)
+ self.beg_str = "sos"
+ self.end_str = "eos"
+ if self.loss_type == "attention":
+ dict_character = [self.beg_str, self.end_str] + dict_character
+ self.dict = {}
+ for i, char in enumerate(dict_character):
+ self.dict[char] = i
+ self.character = dict_character
+
+ def encode(self, text):
+ """convert text-label into text-index.
+ input:
+ text: text labels of each image. [batch_size]
+
+ output:
+ text: concatenated text index for CTCLoss.
+ [sum(text_lengths)] = [text_index_0 + text_index_1 + ... + text_index_(n - 1)]
+ length: length of each text. [batch_size]
+ """
+ if self.character_type == "en":
+ text = text.lower()
+
+ text_list = []
+ for char in text:
+ if char not in self.dict:
+ continue
+ text_list.append(self.dict[char])
+ text = np.array(text_list)
+ return text
+
+ def decode(self, text_index, is_remove_duplicate=False):
+ """ convert text-index into text-label. """
+ char_list = []
+ char_num = self.get_char_num()
+
+ if self.loss_type == "attention":
+ beg_idx = self.get_beg_end_flag_idx("beg")
+ end_idx = self.get_beg_end_flag_idx("end")
+ ignored_tokens = [beg_idx, end_idx]
+ else:
+ ignored_tokens = [char_num]
+
+ for idx in range(len(text_index)):
+ if text_index[idx] in ignored_tokens:
+ continue
+ if is_remove_duplicate:
+ if idx > 0 and text_index[idx - 1] == text_index[idx]:
+ continue
+ char_list.append(self.character[text_index[idx]])
+ text = ''.join(char_list)
+ return text
+
+ def get_char_num(self):
+ return len(self.character)
+
+ def get_beg_end_flag_idx(self, beg_or_end):
+ if self.loss_type == "attention":
+ if beg_or_end == "beg":
+ idx = np.array(self.dict[self.beg_str])
+ elif beg_or_end == "end":
+ idx = np.array(self.dict[self.end_str])
+ else:
+ assert False, "Unsupport type %s in get_beg_end_flag_idx"\
+ % beg_or_end
+ return idx
+ else:
+ err = "error in get_beg_end_flag_idx when using the loss %s"\
+ % (self.loss_type)
+ assert False, err
+
+
+def cal_predicts_accuracy(char_ops,
+ preds,
+ preds_lod,
+ labels,
+ labels_lod,
+ is_remove_duplicate=False):
+ acc_num = 0
+ img_num = 0
+ for ino in range(len(labels_lod) - 1):
+ beg_no = preds_lod[ino]
+ end_no = preds_lod[ino + 1]
+ preds_text = preds[beg_no:end_no].reshape(-1)
+ preds_text = char_ops.decode(preds_text, is_remove_duplicate)
+
+ beg_no = labels_lod[ino]
+ end_no = labels_lod[ino + 1]
+ labels_text = labels[beg_no:end_no].reshape(-1)
+ labels_text = char_ops.decode(labels_text, is_remove_duplicate)
+ img_num += 1
+
+ if preds_text == labels_text:
+ acc_num += 1
+ acc = acc_num * 1.0 / img_num
+ return acc, acc_num, img_num
+
+
+def convert_rec_attention_infer_res(preds):
+ img_num = preds.shape[0]
+ target_lod = [0]
+ convert_ids = []
+ for ino in range(img_num):
+ end_pos = np.where(preds[ino, :] == 1)[0]
+ if len(end_pos) <= 1:
+ text_list = preds[ino, 1:]
+ else:
+ text_list = preds[ino, 1:end_pos[1]]
+ target_lod.append(target_lod[ino] + len(text_list))
+ convert_ids = convert_ids + list(text_list)
+ convert_ids = np.array(convert_ids)
+ convert_ids = convert_ids.reshape((-1, 1))
+ return convert_ids, target_lod
+
+
+def convert_rec_label_to_lod(ori_labels):
+ img_num = len(ori_labels)
+ target_lod = [0]
+ convert_ids = []
+ for ino in range(img_num):
+ target_lod.append(target_lod[ino] + len(ori_labels[ino]))
+ convert_ids = convert_ids + list(ori_labels[ino])
+ convert_ids = np.array(convert_ids)
+ convert_ids = convert_ids.reshape((-1, 1))
+ return convert_ids, target_lod
diff --git a/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/module.py b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/module.py
new file mode 100644
index 0000000000000000000000000000000000000000..6aab2570d30527ae75aa73d2966a2d68a2abc357
--- /dev/null
+++ b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/module.py
@@ -0,0 +1,416 @@
+# -*- coding:utf-8 -*-
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import argparse
+import ast
+import copy
+import math
+import os
+import time
+
+from paddle.fluid.core import AnalysisConfig, create_paddle_predictor, PaddleTensor
+from paddlehub.common.logger import logger
+from paddlehub.module.module import moduleinfo, runnable, serving
+from PIL import Image
+import cv2
+import numpy as np
+import paddle.fluid as fluid
+import paddlehub as hub
+
+from chinese_ocr_db_rcnn.character import CharacterOps
+from chinese_ocr_db_rcnn.utils import draw_ocr, get_image_ext, sorted_boxes
+
+
+@moduleinfo(
+ name="chinese_ocr_db_rcnn",
+ version="1.0.0",
+ summary=
+ "The module can recognize the chinese texts in an image. Firstly, it will detect the text box positions based on the differentiable_binarization_chn module. Then it recognizes the chinese texts. ",
+ author="paddle-dev",
+ author_email="paddle-dev@baidu.com",
+ type="cv/text_recognition")
+class ChineseOCRDBRCNN(hub.Module):
+ def _initialize(self, text_detector_module=None):
+ """
+ initialize with the necessary elements
+ """
+ self.character_dict_path = os.path.join(self.directory, 'assets',
+ 'ppocr_keys_v1.txt')
+ char_ops_params = {
+ 'character_type': 'ch',
+ 'character_dict_path': self.character_dict_path,
+ 'loss_type': 'ctc'
+ }
+ self.char_ops = CharacterOps(char_ops_params)
+ self.rec_image_shape = [3, 32, 320]
+ self._text_detector_module = text_detector_module
+ self.font_file = os.path.join(self.directory, 'assets', 'simfang.ttf')
+ self.pretrained_model_path = os.path.join(self.directory,
+ 'inference_model')
+ self._set_config()
+
+ def _set_config(self):
+ """
+ predictor config setting
+ """
+ model_file_path = os.path.join(self.pretrained_model_path, 'model')
+ params_file_path = os.path.join(self.pretrained_model_path, 'params')
+
+ config = AnalysisConfig(model_file_path, params_file_path)
+ try:
+ _places = os.environ["CUDA_VISIBLE_DEVICES"]
+ int(_places[0])
+ use_gpu = True
+ except:
+ use_gpu = False
+
+ if use_gpu:
+ config.enable_use_gpu(8000, 0)
+ else:
+ config.disable_gpu()
+
+ config.disable_glog_info()
+
+ # use zero copy
+ config.delete_pass("conv_transpose_eltwiseadd_bn_fuse_pass")
+ config.switch_use_feed_fetch_ops(False)
+ self.predictor = create_paddle_predictor(config)
+ input_names = self.predictor.get_input_names()
+ self.input_tensor = self.predictor.get_input_tensor(input_names[0])
+ output_names = self.predictor.get_output_names()
+ self.output_tensors = []
+ for output_name in output_names:
+ output_tensor = self.predictor.get_output_tensor(output_name)
+ self.output_tensors.append(output_tensor)
+
+ @property
+ def text_detector_module(self):
+ """
+ text detect module
+ """
+ if not self._text_detector_module:
+ self._text_detector_module = hub.Module(
+ name='chinese_text_detection_db')
+ return self._text_detector_module
+
+ def read_images(self, paths=[]):
+ images = []
+ for img_path in paths:
+ assert os.path.isfile(
+ img_path), "The {} isn't a valid file.".format(img_path)
+ img = cv2.imread(img_path)
+ if img is None:
+ logger.info("error in loading image:{}".format(img_path))
+ continue
+ images.append(img)
+ return images
+
+ def get_rotate_crop_image(self, img, points):
+ img_height, img_width = img.shape[0:2]
+ left = int(np.min(points[:, 0]))
+ right = int(np.max(points[:, 0]))
+ top = int(np.min(points[:, 1]))
+ bottom = int(np.max(points[:, 1]))
+ img_crop = img[top:bottom, left:right, :].copy()
+ points[:, 0] = points[:, 0] - left
+ points[:, 1] = points[:, 1] - top
+ img_crop_width = int(np.linalg.norm(points[0] - points[1]))
+ img_crop_height = int(np.linalg.norm(points[0] - points[3]))
+ pts_std = np.float32([[0, 0], [img_crop_width, 0],\
+ [img_crop_width, img_crop_height], [0, img_crop_height]])
+ M = cv2.getPerspectiveTransform(points, pts_std)
+ dst_img = cv2.warpPerspective(
+ img_crop,
+ M, (img_crop_width, img_crop_height),
+ borderMode=cv2.BORDER_REPLICATE)
+ dst_img_height, dst_img_width = dst_img.shape[0:2]
+ if dst_img_height * 1.0 / dst_img_width >= 1.5:
+ dst_img = np.rot90(dst_img)
+ return dst_img
+
+ def resize_norm_img(self, img, max_wh_ratio):
+ imgC, imgH, imgW = self.rec_image_shape
+ imgW = int(32 * max_wh_ratio)
+ h = img.shape[0]
+ w = img.shape[1]
+ ratio = w / float(h)
+ if math.ceil(imgH * ratio) > imgW:
+ resized_w = imgW
+ else:
+ resized_w = int(math.ceil(imgH * ratio))
+ resized_image = cv2.resize(img, (resized_w, imgH))
+ resized_image = resized_image.astype('float32')
+ resized_image = resized_image.transpose((2, 0, 1)) / 255
+ resized_image -= 0.5
+ resized_image /= 0.5
+ padding_im = np.zeros((imgC, imgH, imgW), dtype=np.float32)
+ padding_im[:, :, 0:resized_w] = resized_image
+ return padding_im
+
+ @serving
+ def recognize_text(self,
+ images=[],
+ paths=[],
+ use_gpu=False,
+ output_dir='ocr_result',
+ visualization=False,
+ box_thresh=0.5,
+ text_thresh=0.5):
+ """
+ Get the chinese texts in the predicted images.
+ Args:
+ images (list(numpy.ndarray)): images data, shape of each is [H, W, C]. If images not paths
+ paths (list[str]): The paths of images. If paths not images
+ use_gpu (bool): Whether to use gpu.
+ batch_size(int): the program deals once with one
+ output_dir (str): The directory to store output images.
+ visualization (bool): Whether to save image or not.
+ box_thresh(float): the threshold of the detected text box's confidence
+ text_thresh(float): the threshold of the recognize chinese texts' confidence
+ Returns:
+ res (list): The result of chinese texts and save path of images.
+ """
+ if use_gpu:
+ try:
+ _places = os.environ["CUDA_VISIBLE_DEVICES"]
+ int(_places[0])
+ except:
+ raise RuntimeError(
+ "Environment Variable CUDA_VISIBLE_DEVICES is not set correctly. If you wanna use gpu, please set CUDA_VISIBLE_DEVICES via export CUDA_VISIBLE_DEVICES=cuda_device_id."
+ )
+
+ self.use_gpu = use_gpu
+
+ if images != [] and isinstance(images, list) and paths == []:
+ predicted_data = images
+ elif images == [] and isinstance(paths, list) and paths != []:
+ predicted_data = self.read_images(paths)
+ else:
+ raise TypeError("The input data is inconsistent with expectations.")
+
+ assert predicted_data != [], "There is not any image to be predicted. Please check the input data."
+
+ detection_results = self.text_detector_module.detect_text(
+ images=predicted_data, use_gpu=self.use_gpu, box_thresh=box_thresh)
+ boxes = [item['data'] for item in detection_results]
+ all_results = []
+ for index, img_boxes in enumerate(boxes):
+ original_image = predicted_data[index].copy()
+ result = {'save_path': ''}
+ if img_boxes is None:
+ result['data'] = []
+ else:
+ img_crop_list = []
+ boxes = sorted_boxes(img_boxes)
+ for num_box in range(len(boxes)):
+ tmp_box = copy.deepcopy(boxes[num_box])
+ img_crop = self.get_rotate_crop_image(
+ original_image, tmp_box)
+ img_crop_list.append(img_crop)
+
+ rec_results = self._recognize_text(img_crop_list)
+ # if the recognized text confidence score is lower than text_thresh, then drop it
+ rec_res_final = []
+ for index, res in enumerate(rec_results):
+ text, score = res
+ if score >= text_thresh:
+ rec_res_final.append({
+ 'text': text,
+ 'confidence': score,
+ 'text_box_position': boxes[index]
+ })
+ result['data'] = rec_res_final
+
+ if visualization and result['data']:
+ result['save_path'] = self.save_result_image(
+ original_image, boxes, rec_results, output_dir,
+ text_thresh)
+ all_results.append(result)
+
+ return all_results
+
+ def save_result_image(self,
+ original_image,
+ detection_boxes,
+ rec_results,
+ output_dir='ocr_result',
+ text_thresh=0.5):
+ image = Image.fromarray(cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB))
+ txts = [item[0] for item in rec_results]
+ scores = [item[1] for item in rec_results]
+ draw_img = draw_ocr(
+ image,
+ detection_boxes,
+ txts,
+ scores,
+ font_file=self.font_file,
+ draw_txt=True,
+ drop_score=text_thresh)
+
+ if not os.path.exists(output_dir):
+ os.makedirs(output_dir)
+ ext = get_image_ext(original_image)
+ saved_name = 'ndarray_{}{}'.format(time.time(), ext)
+ save_file_path = os.path.join(output_dir, saved_name)
+ cv2.imwrite(save_file_path, draw_img[:, :, ::-1])
+ return save_file_path
+
+ def _recognize_text(self, image_list):
+ img_num = len(image_list)
+ batch_num = 30
+ rec_res = []
+ predict_time = 0
+ for beg_img_no in range(0, img_num, batch_num):
+ end_img_no = min(img_num, beg_img_no + batch_num)
+ norm_img_batch = []
+ max_wh_ratio = 0
+ for ino in range(beg_img_no, end_img_no):
+ h, w = image_list[ino].shape[0:2]
+ wh_ratio = w / h
+ max_wh_ratio = max(max_wh_ratio, wh_ratio)
+ for ino in range(beg_img_no, end_img_no):
+ norm_img = self.resize_norm_img(image_list[ino], max_wh_ratio)
+ norm_img = norm_img[np.newaxis, :]
+ norm_img_batch.append(norm_img)
+ norm_img_batch = np.concatenate(norm_img_batch)
+ norm_img_batch = norm_img_batch.copy()
+ self.input_tensor.copy_from_cpu(norm_img_batch)
+ self.predictor.zero_copy_run()
+ rec_idx_batch = self.output_tensors[0].copy_to_cpu()
+ rec_idx_lod = self.output_tensors[0].lod()[0]
+ predict_batch = self.output_tensors[1].copy_to_cpu()
+ predict_lod = self.output_tensors[1].lod()[0]
+
+ for rno in range(len(rec_idx_lod) - 1):
+ beg = rec_idx_lod[rno]
+ end = rec_idx_lod[rno + 1]
+ rec_idx_tmp = rec_idx_batch[beg:end, 0]
+ preds_text = self.char_ops.decode(rec_idx_tmp)
+ beg = predict_lod[rno]
+ end = predict_lod[rno + 1]
+ probs = predict_batch[beg:end, :]
+ ind = np.argmax(probs, axis=1)
+ blank = probs.shape[1]
+ valid_ind = np.where(ind != (blank - 1))[0]
+ score = np.mean(probs[valid_ind, ind[valid_ind]])
+ rec_res.append([preds_text, score])
+
+ return rec_res
+
+ def save_inference_model(self,
+ dirname,
+ model_filename=None,
+ params_filename=None,
+ combined=True):
+ detector_dir = os.path.join(dirname, 'text_detector')
+ recognizer_dir = os.path.join(dirname, 'text_recognizer')
+ self._save_detector_model(detector_dir, model_filename, params_filename,
+ combined)
+ self._save_recognizer_model(recognizer_dir, model_filename,
+ params_filename, combined)
+ logger.info("The inference model has been saved in the path {}".format(
+ os.path.realpath(dirname)))
+
+ def _save_detector_model(self,
+ dirname,
+ model_filename=None,
+ params_filename=None,
+ combined=True):
+ self.text_detector_module.save_inference_model(
+ dirname, model_filename, params_filename, combined)
+
+ def _save_recognizer_model(self,
+ dirname,
+ model_filename=None,
+ params_filename=None,
+ combined=True):
+ if combined:
+ model_filename = "__model__" if not model_filename else model_filename
+ params_filename = "__params__" if not params_filename else params_filename
+ place = fluid.CPUPlace()
+ exe = fluid.Executor(place)
+
+ model_file_path = os.path.join(self.pretrained_model_path, 'model')
+ params_file_path = os.path.join(self.pretrained_model_path, 'params')
+ program, feeded_var_names, target_vars = fluid.io.load_inference_model(
+ dirname=self.pretrained_model_path,
+ model_filename=model_file_path,
+ params_filename=params_file_path,
+ executor=exe)
+
+ fluid.io.save_inference_model(
+ dirname=dirname,
+ main_program=program,
+ executor=exe,
+ feeded_var_names=feeded_var_names,
+ target_vars=target_vars,
+ model_filename=model_filename,
+ params_filename=params_filename)
+
+ @runnable
+ def run_cmd(self, argvs):
+ """
+ Run as a command
+ """
+ self.parser = argparse.ArgumentParser(
+ description="Run the chinese_ocr_db_rcnn module.",
+ prog='hub run chinese_ocr_db_rcnn',
+ usage='%(prog)s',
+ add_help=True)
+
+ self.arg_input_group = self.parser.add_argument_group(
+ title="Input options", description="Input data. Required")
+ self.arg_config_group = self.parser.add_argument_group(
+ title="Config options",
+ description=
+ "Run configuration for controlling module behavior, not required.")
+
+ self.add_module_config_arg()
+ self.add_module_input_arg()
+
+ args = self.parser.parse_args(argvs)
+ results = self.recognize_texts(
+ paths=[args.input_path],
+ use_gpu=args.use_gpu,
+ output_dir=args.output_dir,
+ visualization=args.visualization)
+ return results
+
+ def add_module_config_arg(self):
+ """
+ Add the command config options
+ """
+ self.arg_config_group.add_argument(
+ '--use_gpu',
+ type=ast.literal_eval,
+ default=False,
+ help="whether use GPU or not")
+ self.arg_config_group.add_argument(
+ '--output_dir',
+ type=str,
+ default='ocr_result',
+ help="The directory to save output images.")
+ self.arg_config_group.add_argument(
+ '--visualization',
+ type=ast.literal_eval,
+ default=False,
+ help="whether to save output as images.")
+
+ def add_module_input_arg(self):
+ """
+ Add the command input options
+ """
+ self.arg_input_group.add_argument(
+ '--input_path', type=str, default=None, help="diretory to image")
+
+
+if __name__ == '__main__':
+ ocr = ChineseOCRDBRCNN()
+ image_path = [
+ '../doc/imgs/11.jpg', '../doc/imgs/12.jpg', '../test_image.jpg'
+ ]
+ res = ocr.recognize_text(paths=image_path, visualization=True)
+ ocr.save_inference_model('save')
+ print(res)
diff --git a/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/utils.py b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/utils.py
new file mode 100644
index 0000000000000000000000000000000000000000..0a8574fbe16bc29a8cf283381fc96f81007e7ddd
--- /dev/null
+++ b/hub_module/modules/image/text_recognition/chinese_ocr_db_rcnn/utils.py
@@ -0,0 +1,105 @@
+# -*- coding:utf-8 -*-
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+from PIL import Image, ImageDraw, ImageFont
+import cv2
+import numpy as np
+
+
+def draw_ocr(image,
+ boxes,
+ txts,
+ scores,
+ font_file,
+ draw_txt=True,
+ drop_score=0.5):
+ img = image.copy()
+ draw = ImageDraw.Draw(img)
+ if scores is None:
+ scores = [1] * len(boxes)
+ for (box, score) in zip(boxes, scores):
+ if score < drop_score:
+ continue
+ draw.line([(box[0][0], box[0][1]), (box[1][0], box[1][1])], fill='red')
+ draw.line([(box[1][0], box[1][1]), (box[2][0], box[2][1])], fill='red')
+ draw.line([(box[2][0], box[2][1]), (box[3][0], box[3][1])], fill='red')
+ draw.line([(box[3][0], box[3][1]), (box[0][0], box[0][1])], fill='red')
+ draw.line([(box[0][0] - 1, box[0][1] + 1),
+ (box[1][0] - 1, box[1][1] + 1)],
+ fill='red')
+ draw.line([(box[1][0] - 1, box[1][1] + 1),
+ (box[2][0] - 1, box[2][1] + 1)],
+ fill='red')
+ draw.line([(box[2][0] - 1, box[2][1] + 1),
+ (box[3][0] - 1, box[3][1] + 1)],
+ fill='red')
+ draw.line([(box[3][0] - 1, box[3][1] + 1),
+ (box[0][0] - 1, box[0][1] + 1)],
+ fill='red')
+
+ if draw_txt:
+ txt_color = (0, 0, 0)
+ img = np.array(resize_img(img))
+ _h = img.shape[0]
+ blank_img = np.ones(shape=[_h, 600], dtype=np.int8) * 255
+ blank_img = Image.fromarray(blank_img).convert("RGB")
+ draw_txt = ImageDraw.Draw(blank_img)
+
+ font_size = 20
+ gap = 20
+ title = "index text score"
+ font = ImageFont.truetype(font_file, font_size, encoding="utf-8")
+
+ draw_txt.text((20, 0), title, txt_color, font=font)
+ count = 0
+ for idx, txt in enumerate(txts):
+ if scores[idx] < drop_score:
+ continue
+ font = ImageFont.truetype(font_file, font_size, encoding="utf-8")
+ new_txt = str(count) + ': ' + txt + ' ' + str(scores[count])
+ draw_txt.text((20, gap * (count + 1)),
+ new_txt,
+ txt_color,
+ font=font)
+ count += 1
+ img = np.concatenate([np.array(img), np.array(blank_img)], axis=1)
+ return img
+
+
+def resize_img(img, input_size=600):
+ img = np.array(img)
+ im_shape = img.shape
+ im_size_min = np.min(im_shape[0:2])
+ im_size_max = np.max(im_shape[0:2])
+ im_scale = float(input_size) / float(im_size_max)
+ im = cv2.resize(img, None, None, fx=im_scale, fy=im_scale)
+ return im
+
+
+def get_image_ext(image):
+ if image.shape[2] == 4:
+ return ".png"
+ return ".jpg"
+
+
+def sorted_boxes(dt_boxes):
+ """
+ Sort text boxes in order from top to bottom, left to right
+ args:
+ dt_boxes(array):detected text boxes with shape [4, 2]
+ return:
+ sorted boxes(array) with shape [4, 2]
+ """
+ num_boxes = dt_boxes.shape[0]
+ sorted_boxes = sorted(dt_boxes, key=lambda x: x[0][1])
+ _boxes = list(sorted_boxes)
+
+ for i in range(num_boxes - 1):
+ if abs(_boxes[i+1][0][1] - _boxes[i][0][1]) < 10 and \
+ (_boxes[i + 1][0][0] < _boxes[i][0][0]):
+ tmp = _boxes[i]
+ _boxes[i] = _boxes[i + 1]
+ _boxes[i + 1] = tmp
+ return _boxes
diff --git a/hub_module/modules/image/text_recognition/chinese_text_detection_db/README.md b/hub_module/modules/image/text_recognition/chinese_text_detection_db/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..71ac3537de1660cb844b1c5588a7ffbeca4b6b62
--- /dev/null
+++ b/hub_module/modules/image/text_recognition/chinese_text_detection_db/README.md
@@ -0,0 +1,119 @@
+## 概述
+
+Differentiable Binarization（简称DB）是一种基于分割的文本检测算法。在各种文本检测算法中，基于分割的检测算法可以更好地处理弯曲等不规则形状文本，因此往往能取得更好的检测效果。但分割法后处理步骤中将分割结果转化为检测框的流程复杂，耗时严重。DB将二值化阈值加入训练中学习，可以获得更准确的检测边界，从而简化后处理流程。该Module支持直接预测。
+
+
+
+
+
+更多详情参考[Real-time Scene Text Detection with Differentiable Binarization](https://arxiv.org/pdf/1911.08947.pdf)
+
+
+## 命令行预测
+
+```shell
+$ hub run chinese_text_detection_db --input_path "/PATH/TO/IMAGE"
+```
+
+**该Module依赖于第三方库shapely和pyclipper，使用该Module之前，请先安装shapely和pyclipper。**
+
+## API
+
+```python
+def detect_text(paths=[],
+ images=[],
+ use_gpu=False,
+ output_dir='detection_result',
+ box_thresh=0.5,
+ visualization=False)
+```
+
+预测API，检测输入图片中的所有中文文本的位置。
+
+**参数**
+
+* paths (list\[str\]): 图片的路径；
+* images (list\[numpy.ndarray\]): 图片数据，ndarray.shape 为 \[H, W, C\]，BGR格式；
+* use\_gpu (bool): 是否使用 GPU；**若使用GPU，请先设置CUDA_VISIBLE_DEVICES环境变量**
+* box\_thresh (float): 检测文本框置信度的阈值；
+* visualization (bool): 是否将识别结果保存为图片文件；
+* output\_dir (str): 图片的保存路径，默认设为 detection\_result；
+
+**返回**
+
+* res (list\[dict\]): 识别结果的列表，列表中每一个元素为 dict，各字段为：
+ * data (list): 检测文本框结果，numpy.ndarray，文本框在原图中的像素坐标，4*2的矩阵，依次表示文本框左下、右下、右上、左上顶点的坐标
+ * save_path (str): 识别结果的保存路径, 如不保存图片则save_path为''
+
+### 代码示例
+
+```python
+import paddlehub as hub
+import cv2
+
+text_detector = hub.Module(name="chinese_text_detection_db")
+result = text_detector.detect_text(images=[cv2.imread('/PATH/TO/IMAGE')])
+
+# or
+# result =text_detector.detect_text(paths=['/PATH/TO/IMAGE'])
+```
+
+
+## 服务部署
+
+PaddleHub Serving 可以部署一个目标检测的在线服务。
+
+### 第一步：启动PaddleHub Serving
+
+运行启动命令：
+```shell
+$ hub serving start -m chinese_text_detection_db
+```
+
+这样就完成了一个目标检测的服务化API的部署，默认端口号为8866。
+
+**NOTE:** 如使用GPU预测，则需要在启动服务之前，请设置CUDA\_VISIBLE\_DEVICES环境变量，否则不用设置。
+
+### 第二步：发送预测请求
+
+配置好服务端，以下数行代码即可实现发送预测请求，获取预测结果
+
+```python
+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_text_detection_db"
+r = requests.post(url=url, headers=headers, data=json.dumps(data))
+
+# 打印预测结果
+print(r.json()["results"])
+```
+
+## 查看代码
+
+https://github.com/PaddlePaddle/PaddleOCR
+
+## 依赖
+
+paddlepaddle >= 1.7.2
+
+paddlehub >= 1.6.0
+
+shapely
+
+pyclipper
+
+## 更新历史
+
+* 1.0.0
+
+ 初始发布
diff --git a/hub_module/modules/image/text_recognition/chinese_text_detection_db/__init__.py b/hub_module/modules/image/text_recognition/chinese_text_detection_db/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/hub_module/modules/image/text_recognition/chinese_text_detection_db/module.py b/hub_module/modules/image/text_recognition/chinese_text_detection_db/module.py
new file mode 100644
index 0000000000000000000000000000000000000000..c2df6afeb7d104bc3073c73dde3f2f6260fc34f0
--- /dev/null
+++ b/hub_module/modules/image/text_recognition/chinese_text_detection_db/module.py
@@ -0,0 +1,313 @@
+# -*- coding:utf-8 -*-
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import argparse
+import ast
+import math
+import os
+import time
+
+from paddle.fluid.core import AnalysisConfig, create_paddle_predictor, PaddleTensor
+from paddlehub.common.logger import logger
+from paddlehub.module.module import moduleinfo, runnable, serving
+from PIL import Image
+import cv2
+import numpy as np
+import paddle.fluid as fluid
+import paddlehub as hub
+
+from chinese_text_detection_db.processor import DBPreProcess, DBPostProcess, draw_boxes, get_image_ext
+
+
+@moduleinfo(
+ name="chinese_text_detection_db",
+ version="1.0.0",
+ summary=
+ "The module aims to detect chinese text position in the image, which is based on differentiable_binarization algorithm.",
+ author="paddle-dev",
+ author_email="paddle-dev@baidu.com",
+ type="cv/text_recognition")
+class ChineseTextDetectionDB(hub.Module):
+ def _initialize(self):
+ """
+ initialize with the necessary elements
+ """
+ self.check_requirements()
+ self.pretrained_model_path = os.path.join(self.directory,
+ 'inference_model')
+ self._set_config()
+
+ def _set_config(self):
+ """
+ predictor config setting
+ """
+ model_file_path = os.path.join(self.pretrained_model_path, 'model')
+ params_file_path = os.path.join(self.pretrained_model_path, 'params')
+
+ config = AnalysisConfig(model_file_path, params_file_path)
+ try:
+ _places = os.environ["CUDA_VISIBLE_DEVICES"]
+ int(_places[0])
+ use_gpu = True
+ except:
+ use_gpu = False
+
+ if use_gpu:
+ config.enable_use_gpu(8000, 0)
+ else:
+ config.disable_gpu()
+
+ config.disable_glog_info()
+
+ # use zero copy
+ config.delete_pass("conv_transpose_eltwiseadd_bn_fuse_pass")
+ config.switch_use_feed_fetch_ops(False)
+ self.predictor = create_paddle_predictor(config)
+ input_names = self.predictor.get_input_names()
+ self.input_tensor = self.predictor.get_input_tensor(input_names[0])
+ output_names = self.predictor.get_output_names()
+ self.output_tensors = []
+ for output_name in output_names:
+ output_tensor = self.predictor.get_output_tensor(output_name)
+ self.output_tensors.append(output_tensor)
+
+ def check_requirements(self):
+ try:
+ import shapely, pyclipper
+ except:
+ logger.error(
+ 'This module requires the shapely, pyclipper tools. The running enviroment does not meet the requirments. Please install the two packages.'
+ )
+ exit()
+
+ def read_images(self, paths=[]):
+ images = []
+ for img_path in paths:
+ assert os.path.isfile(
+ img_path), "The {} isn't a valid file.".format(img_path)
+ img = cv2.imread(img_path)
+ if img is None:
+ logger.info("error in loading image:{}".format(img_path))
+ continue
+ images.append(img)
+ return images
+
+ def filter_tag_det_res(self, dt_boxes, image_shape):
+ img_height, img_width = image_shape[0:2]
+ dt_boxes_new = []
+ for box in dt_boxes:
+ box = self.order_points_clockwise(box)
+ left = int(np.min(box[:, 0]))
+ right = int(np.max(box[:, 0]))
+ top = int(np.min(box[:, 1]))
+ bottom = int(np.max(box[:, 1]))
+ bbox_height = bottom - top
+ bbox_width = right - left
+ diffh = math.fabs(box[0, 1] - box[1, 1])
+ diffw = math.fabs(box[0, 0] - box[3, 0])
+ rect_width = int(np.linalg.norm(box[0] - box[1]))
+ rect_height = int(np.linalg.norm(box[0] - box[3]))
+ if rect_width <= 10 or rect_height <= 10:
+ continue
+ dt_boxes_new.append(box)
+ dt_boxes = np.array(dt_boxes_new)
+ return dt_boxes
+
+ def order_points_clockwise(self, pts):
+ """
+ reference from: https://github.com/jrosebr1/imutils/blob/master/imutils/perspective.py
+ # sort the points based on their x-coordinates
+ """
+ xSorted = pts[np.argsort(pts[:, 0]), :]
+
+ # grab the left-most and right-most points from the sorted
+ # x-roodinate points
+ leftMost = xSorted[:2, :]
+ rightMost = xSorted[2:, :]
+
+ # now, sort the left-most coordinates according to their
+ # y-coordinates so we can grab the top-left and bottom-left
+ # points, respectively
+ leftMost = leftMost[np.argsort(leftMost[:, 1]), :]
+ (tl, bl) = leftMost
+
+ rightMost = rightMost[np.argsort(rightMost[:, 1]), :]
+ (tr, br) = rightMost
+
+ rect = np.array([tl, tr, br, bl], dtype="float32")
+ return rect
+
+ @serving
+ def detect_text(self,
+ images=[],
+ paths=[],
+ use_gpu=False,
+ output_dir='detection_result',
+ visualization=False,
+ box_thresh=0.5):
+ """
+ Get the text box in the predicted images.
+ Args:
+ images (list(numpy.ndarray)): images data, shape of each is [H, W, C]. If images not paths
+ paths (list[str]): The paths of images. If paths not images
+ use_gpu (bool): Whether to use gpu. Default false.
+ output_dir (str): The directory to store output images.
+ visualization (bool): Whether to save image or not.
+ box_thresh(float): the threshold of the detected text box's confidence
+ Returns:
+ res (list): The result of text detection box and save path of images.
+ """
+ if use_gpu:
+ try:
+ _places = os.environ["CUDA_VISIBLE_DEVICES"]
+ int(_places[0])
+ except:
+ raise RuntimeError(
+ "Environment Variable CUDA_VISIBLE_DEVICES is not set correctly. If you wanna use gpu, please set CUDA_VISIBLE_DEVICES via export CUDA_VISIBLE_DEVICES=cuda_device_id."
+ )
+
+ if images != [] and isinstance(images, list) and paths == []:
+ predicted_data = images
+ elif images == [] and isinstance(paths, list) and paths != []:
+ predicted_data = self.read_images(paths)
+ else:
+ raise TypeError("The input data is inconsistent with expectations.")
+
+ assert predicted_data != [], "There is not any image to be predicted. Please check the input data."
+
+ preprocessor = DBPreProcess()
+ postprocessor = DBPostProcess(box_thresh)
+
+ all_imgs = []
+ all_ratios = []
+ all_results = []
+ for original_image in predicted_data:
+ im, ratio_list = preprocessor(original_image)
+ res = {'save_path': ''}
+ if im is None:
+ res['data'] = []
+
+ else:
+ im = im.copy()
+ starttime = time.time()
+ self.input_tensor.copy_from_cpu(im)
+ self.predictor.zero_copy_run()
+ data_out = self.output_tensors[0].copy_to_cpu()
+ dt_boxes_list = postprocessor(data_out, [ratio_list])
+ boxes = self.filter_tag_det_res(dt_boxes_list[0],
+ original_image.shape)
+ res['data'] = boxes
+
+ all_imgs.append(im)
+ all_ratios.append(ratio_list)
+ if visualization:
+ img = Image.fromarray(
+ cv2.cvtColor(original_image, cv2.COLOR_BGR2RGB))
+ draw_img = draw_boxes(img, boxes)
+ draw_img = np.array(draw_img)
+ if not os.path.exists(output_dir):
+ os.makedirs(output_dir)
+ ext = get_image_ext(original_image)
+ saved_name = 'ndarray_{}{}'.format(time.time(), ext)
+ cv2.imwrite(
+ os.path.join(output_dir, saved_name),
+ draw_img[:, :, ::-1])
+ res['save_path'] = os.path.join(output_dir, saved_name)
+
+ all_results.append(res)
+
+ return all_results
+
+ def save_inference_model(self,
+ dirname,
+ model_filename=None,
+ params_filename=None,
+ combined=True):
+ if combined:
+ model_filename = "__model__" if not model_filename else model_filename
+ params_filename = "__params__" if not params_filename else params_filename
+ place = fluid.CPUPlace()
+ exe = fluid.Executor(place)
+
+ model_file_path = os.path.join(self.pretrained_model_path, 'model')
+ params_file_path = os.path.join(self.pretrained_model_path, 'params')
+ program, feeded_var_names, target_vars = fluid.io.load_inference_model(
+ dirname=self.pretrained_model_path,
+ model_filename=model_file_path,
+ params_filename=params_file_path,
+ executor=exe)
+
+ fluid.io.save_inference_model(
+ dirname=dirname,
+ main_program=program,
+ executor=exe,
+ feeded_var_names=feeded_var_names,
+ target_vars=target_vars,
+ model_filename=model_filename,
+ params_filename=params_filename)
+
+ @runnable
+ def run_cmd(self, argvs):
+ """
+ Run as a command
+ """
+ self.parser = argparse.ArgumentParser(
+ description="Run the chinese_text_detection_db module.",
+ prog='hub run chinese_text_detection_db',
+ usage='%(prog)s',
+ add_help=True)
+
+ self.arg_input_group = self.parser.add_argument_group(
+ title="Input options", description="Input data. Required")
+ self.arg_config_group = self.parser.add_argument_group(
+ title="Config options",
+ description=
+ "Run configuration for controlling module behavior, not required.")
+
+ self.add_module_config_arg()
+ self.add_module_input_arg()
+
+ args = self.parser.parse_args(argvs)
+ results = self.detect_text(
+ paths=[args.input_path],
+ use_gpu=args.use_gpu,
+ output_dir=args.output_dir,
+ visualization=args.visualization)
+ return results
+
+ def add_module_config_arg(self):
+ """
+ Add the command config options
+ """
+ self.arg_config_group.add_argument(
+ '--use_gpu',
+ type=ast.literal_eval,
+ default=False,
+ help="whether use GPU or not")
+ self.arg_config_group.add_argument(
+ '--output_dir',
+ type=str,
+ default='detection_result',
+ help="The directory to save output images.")
+ self.arg_config_group.add_argument(
+ '--visualization',
+ type=ast.literal_eval,
+ default=False,
+ help="whether to save output as images.")
+
+ def add_module_input_arg(self):
+ """
+ Add the command input options
+ """
+ self.arg_input_group.add_argument(
+ '--input_path', type=str, default=None, help="diretory to image")
+
+
+if __name__ == '__main__':
+ db = ChineseTextDetectionDB()
+ image_path = ['../doc/imgs/11.jpg', '../doc/imgs/12.jpg']
+ res = db.detect_text(paths=image_path, visualization=True)
+ db.save_inference_model('save')
+ print(res)
diff --git a/hub_module/modules/image/text_recognition/chinese_text_detection_db/processor.py b/hub_module/modules/image/text_recognition/chinese_text_detection_db/processor.py
new file mode 100644
index 0000000000000000000000000000000000000000..aec5a11953bc094e21401acb81ca0074e22fd5de
--- /dev/null
+++ b/hub_module/modules/image/text_recognition/chinese_text_detection_db/processor.py
@@ -0,0 +1,237 @@
+# -*- coding:utf-8 -*-
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import sys
+
+from PIL import Image, ImageDraw, ImageFont
+from shapely.geometry import Polygon
+import cv2
+import numpy as np
+import pyclipper
+
+
+class DBPreProcess(object):
+ def __init__(self, max_side_len=960):
+ self.max_side_len = max_side_len
+
+ def resize_image_type(self, im):
+ """
+ resize image to a size multiple of 32 which is required by the network
+ """
+ h, w, _ = im.shape
+
+ resize_w = w
+ resize_h = h
+
+ # limit the max side
+ if max(resize_h, resize_w) > self.max_side_len:
+ if resize_h > resize_w:
+ ratio = float(self.max_side_len) / resize_h
+ else:
+ ratio = float(self.max_side_len) / resize_w
+ else:
+ ratio = 1.
+ resize_h = int(resize_h * ratio)
+ resize_w = int(resize_w * ratio)
+ if resize_h % 32 == 0:
+ resize_h = resize_h
+ elif resize_h // 32 <= 1:
+ resize_h = 32
+ else:
+ resize_h = (resize_h // 32 - 1) * 32
+ if resize_w % 32 == 0:
+ resize_w = resize_w
+ elif resize_w // 32 <= 1:
+ resize_w = 32
+ else:
+ resize_w = (resize_w // 32 - 1) * 32
+ try:
+ if int(resize_w) <= 0 or int(resize_h) <= 0:
+ return None, (None, None)
+ im = cv2.resize(im, (int(resize_w), int(resize_h)))
+ except:
+ print(im.shape, resize_w, resize_h)
+ sys.exit(0)
+ ratio_h = resize_h / float(h)
+ ratio_w = resize_w / float(w)
+ return im, (ratio_h, ratio_w)
+
+ def normalize(self, im):
+ img_mean = [0.485, 0.456, 0.406]
+ img_std = [0.229, 0.224, 0.225]
+ im = im.astype(np.float32, copy=False)
+ im = im / 255
+ im -= img_mean
+ im /= img_std
+ channel_swap = (2, 0, 1)
+ im = im.transpose(channel_swap)
+ return im
+
+ def __call__(self, im):
+ im, (ratio_h, ratio_w) = self.resize_image_type(im)
+ im = self.normalize(im)
+ im = im[np.newaxis, :]
+ return [im, (ratio_h, ratio_w)]
+
+
+class DBPostProcess(object):
+ """
+ The post process for Differentiable Binarization (DB).
+ """
+
+ def __init__(self, thresh=0.3, box_thresh=0.5, max_candidates=1000):
+ self.thresh = thresh
+ self.box_thresh = box_thresh
+ self.max_candidates = max_candidates
+ self.min_size = 3
+
+ def boxes_from_bitmap(self, pred, _bitmap, dest_width, dest_height):
+ '''
+ _bitmap: single map with shape (1, H, W),
+ whose values are binarized as {0, 1}
+ '''
+
+ bitmap = _bitmap
+ height, width = bitmap.shape
+
+ outs = cv2.findContours((bitmap * 255).astype(np.uint8), cv2.RETR_LIST,
+ cv2.CHAIN_APPROX_SIMPLE)
+ if len(outs) == 3:
+ img, contours, _ = outs[0], outs[1], outs[2]
+ elif len(outs) == 2:
+ contours, _ = outs[0], outs[1]
+
+ num_contours = min(len(contours), self.max_candidates)
+ boxes = np.zeros((num_contours, 4, 2), dtype=np.int16)
+ scores = np.zeros((num_contours, ), dtype=np.float32)
+
+ for index in range(num_contours):
+ contour = contours[index]
+ points, sside = self.get_mini_boxes(contour)
+ if sside < self.min_size:
+ continue
+ points = np.array(points)
+ score = self.box_score_fast(pred, points.reshape(-1, 2))
+ if self.box_thresh > score:
+ continue
+
+ box = self.unclip(points).reshape(-1, 1, 2)
+ box, sside = self.get_mini_boxes(box)
+ if sside < self.min_size + 2:
+ continue
+ box = np.array(box)
+ if not isinstance(dest_width, int):
+ dest_width = dest_width.item()
+ dest_height = dest_height.item()
+
+ box[:, 0] = np.clip(
+ np.round(box[:, 0] / width * dest_width), 0, dest_width)
+ box[:, 1] = np.clip(
+ np.round(box[:, 1] / height * dest_height), 0, dest_height)
+ boxes[index, :, :] = box.astype(np.int16)
+ scores[index] = score
+ return boxes, scores
+
+ def unclip(self, box, unclip_ratio=2.0):
+ poly = Polygon(box)
+ distance = poly.area * unclip_ratio / poly.length
+ offset = pyclipper.PyclipperOffset()
+ offset.AddPath(box, pyclipper.JT_ROUND, pyclipper.ET_CLOSEDPOLYGON)
+ expanded = np.array(offset.Execute(distance))
+ return expanded
+
+ def get_mini_boxes(self, contour):
+ bounding_box = cv2.minAreaRect(contour)
+ points = sorted(list(cv2.boxPoints(bounding_box)), key=lambda x: x[0])
+
+ index_1, index_2, index_3, index_4 = 0, 1, 2, 3
+ if points[1][1] > points[0][1]:
+ index_1 = 0
+ index_4 = 1
+ else:
+ index_1 = 1
+ index_4 = 0
+ if points[3][1] > points[2][1]:
+ index_2 = 2
+ index_3 = 3
+ else:
+ index_2 = 3
+ index_3 = 2
+
+ box = [
+ points[index_1], points[index_2], points[index_3], points[index_4]
+ ]
+ return box, min(bounding_box[1])
+
+ def box_score_fast(self, bitmap, _box):
+ h, w = bitmap.shape[:2]
+ box = _box.copy()
+ xmin = np.clip(np.floor(box[:, 0].min()).astype(np.int), 0, w - 1)
+ xmax = np.clip(np.ceil(box[:, 0].max()).astype(np.int), 0, w - 1)
+ ymin = np.clip(np.floor(box[:, 1].min()).astype(np.int), 0, h - 1)
+ ymax = np.clip(np.ceil(box[:, 1].max()).astype(np.int), 0, h - 1)
+
+ mask = np.zeros((ymax - ymin + 1, xmax - xmin + 1), dtype=np.uint8)
+ box[:, 0] = box[:, 0] - xmin
+ box[:, 1] = box[:, 1] - ymin
+ cv2.fillPoly(mask, box.reshape(1, -1, 2).astype(np.int32), 1)
+ return cv2.mean(bitmap[ymin:ymax + 1, xmin:xmax + 1], mask)[0]
+
+ def __call__(self, predictions, ratio_list):
+ pred = predictions[:, 0, :, :]
+ segmentation = pred > self.thresh
+
+ boxes_batch = []
+ for batch_index in range(pred.shape[0]):
+ height, width = pred.shape[-2:]
+ tmp_boxes, tmp_scores = self.boxes_from_bitmap(
+ pred[batch_index], segmentation[batch_index], width, height)
+
+ boxes = []
+ for k in range(len(tmp_boxes)):
+ if tmp_scores[k] > self.box_thresh:
+ boxes.append(tmp_boxes[k])
+ if len(boxes) > 0:
+ boxes = np.array(boxes)
+
+ ratio_h, ratio_w = ratio_list[batch_index]
+ boxes[:, :, 0] = boxes[:, :, 0] / ratio_w
+ boxes[:, :, 1] = boxes[:, :, 1] / ratio_h
+
+ boxes_batch.append(boxes)
+ return boxes_batch
+
+
+def draw_boxes(image, boxes, scores=None, drop_score=0.5):
+ img = image.copy()
+ draw = ImageDraw.Draw(img)
+ if scores is None:
+ scores = [1] * len(boxes)
+ for (box, score) in zip(boxes, scores):
+ if score < drop_score:
+ continue
+ draw.line([(box[0][0], box[0][1]), (box[1][0], box[1][1])], fill='red')
+ draw.line([(box[1][0], box[1][1]), (box[2][0], box[2][1])], fill='red')
+ draw.line([(box[2][0], box[2][1]), (box[3][0], box[3][1])], fill='red')
+ draw.line([(box[3][0], box[3][1]), (box[0][0], box[0][1])], fill='red')
+ draw.line([(box[0][0] - 1, box[0][1] + 1),
+ (box[1][0] - 1, box[1][1] + 1)],
+ fill='red')
+ draw.line([(box[1][0] - 1, box[1][1] + 1),
+ (box[2][0] - 1, box[2][1] + 1)],
+ fill='red')
+ draw.line([(box[2][0] - 1, box[2][1] + 1),
+ (box[3][0] - 1, box[3][1] + 1)],
+ fill='red')
+ draw.line([(box[3][0] - 1, box[3][1] + 1),
+ (box[0][0] - 1, box[0][1] + 1)],
+ fill='red')
+ return img
+
+
+def get_image_ext(image):
+ if image.shape[2] == 4:
+ return ".png"
+ return ".jpg"
diff --git a/hub_module/scripts/configs/chinese_ocr_db_rcnn.yml b/hub_module/scripts/configs/chinese_ocr_db_rcnn.yml
new file mode 100644
index 0000000000000000000000000000000000000000..a50b75c9672dda57133bcf8c19979ef14fd26aa6
--- /dev/null
+++ b/hub_module/scripts/configs/chinese_ocr_db_rcnn.yml
@@ -0,0 +1,10 @@
+name: chinese_ocr_db_rcnn
+dir: "modules/image/text_recognition/chinese_ocr_db_rcnn"
+exclude:
+ - README.md
+
+resources:
+ -
+ url: https://bj.bcebos.com/paddlehub/model/image/ocr/chinese_ocr_db_rcnn_infer_model.tar.gz
+ dest: .
+ uncompress: True
diff --git a/hub_module/scripts/configs/chinese_text_detection_db.yml b/hub_module/scripts/configs/chinese_text_detection_db.yml
new file mode 100644
index 0000000000000000000000000000000000000000..20b07deffcf2e33e49d5b5411be9785804a932d7
--- /dev/null
+++ b/hub_module/scripts/configs/chinese_text_detection_db.yml
@@ -0,0 +1,10 @@
+name: chinese_text_detection_db
+dir: "modules/image/text_recognition/chinese_text_detection_db"
+exclude:
+ - README.md
+
+resources:
+ -
+ url: https://bj.bcebos.com/paddlehub/model/image/ocr/chinese_text_detection_db_infer_model.tar.gz
+ dest: .
+ uncompress: True
diff --git a/hub_module/tests/image_dataset/text_recognition/11.jpg b/hub_module/tests/image_dataset/text_recognition/11.jpg
new file mode 100755
index 0000000000000000000000000000000000000000..ed91b8c5ca2a348fe7b138e83114ff81ecb107de
Binary files /dev/null and b/hub_module/tests/image_dataset/text_recognition/11.jpg differ
diff --git a/hub_module/tests/image_dataset/text_recognition/test_image.jpg b/hub_module/tests/image_dataset/text_recognition/test_image.jpg
new file mode 100644
index 0000000000000000000000000000000000000000..be103f39ec5c2a4e4681ffb82bf8231feef1c048
Binary files /dev/null and b/hub_module/tests/image_dataset/text_recognition/test_image.jpg differ
diff --git a/hub_module/tests/unittests/test_chinese_ocr_db_rcnn.py b/hub_module/tests/unittests/test_chinese_ocr_db_rcnn.py
new file mode 100644
index 0000000000000000000000000000000000000000..89d35abb14c6c87fe5620d8b49f6b97ff6087b9b
--- /dev/null
+++ b/hub_module/tests/unittests/test_chinese_ocr_db_rcnn.py
@@ -0,0 +1,58 @@
+# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import os
+from unittest import TestCase, main
+os.environ['CUDA_VISIBLE_DEVICES'] = '0'
+
+import cv2
+import paddlehub as hub
+
+
+class ChineseOCRDBRCNNTestCase(TestCase):
+ def setUp(self):
+ self.module = hub.Module(name='chinese_ocr_db_rcnn')
+ self.test_images = [
+ "../image_dataset/text_recognition/11.jpg",
+ "../image_dataset/text_recognition/test_image.jpg"
+ ]
+
+ def test_detect_text(self):
+ results_1 = self.module.recognize_text(
+ paths=self.test_images, use_gpu=True)
+ results_2 = self.module.recognize_text(
+ paths=self.test_images, use_gpu=False)
+
+ test_images = [cv2.imread(img) for img in self.test_images]
+ results_3 = self.module.recognize_text(
+ images=test_images, use_gpu=False)
+ for i, res in enumerate(results_1):
+ self.assertEqual(res['save_path'], '')
+
+ for j, item in enumerate(res['data']):
+ self.assertEqual(item['confidence'],
+ results_2[i]['data'][j]['confidence'])
+ self.assertEqual(item['confidence'],
+ results_3[i]['data'][j]['confidence'])
+ self.assertEqual(item['text'], results_2[i]['data'][j]['text'])
+ self.assertEqual(item['text'], results_3[i]['data'][j]['text'])
+ self.assertEqual(
+ (item['text_box_position'].all() == results_2[i]['data'][j]
+ ['text_box_position'].all()), True)
+ self.assertEqual(
+ (item['text_box_position'].all() == results_3[i]['data'][j]
+ ['text_box_position'].all()), True)
+
+
+if __name__ == '__main__':
+ main()
diff --git a/hub_module/tests/unittests/test_chinese_text_detection_db.py b/hub_module/tests/unittests/test_chinese_text_detection_db.py
new file mode 100644
index 0000000000000000000000000000000000000000..cb6fa5006ad678c1685e6b7bd068b4c8561fb844
--- /dev/null
+++ b/hub_module/tests/unittests/test_chinese_text_detection_db.py
@@ -0,0 +1,47 @@
+# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+import os
+from unittest import TestCase, main
+os.environ['CUDA_VISIBLE_DEVICES'] = '0'
+
+import cv2
+import paddlehub as hub
+
+
+class ChineseTextDetectionDBTestCase(TestCase):
+ def setUp(self):
+ self.module = hub.Module(name='chinese_text_detection_db')
+ self.test_images = [
+ "../image_dataset/text_recognition/11.jpg",
+ "../image_dataset/text_recognition/test_image.jpg"
+ ]
+
+ def test_detect_text(self):
+ results_1 = self.module.detect_text(
+ paths=self.test_images, use_gpu=True)
+ results_2 = self.module.detect_text(
+ paths=self.test_images, use_gpu=False)
+
+ test_images = [cv2.imread(img) for img in self.test_images]
+ results_3 = self.module.detect_text(images=test_images, use_gpu=False)
+ for index, res in enumerate(results_1):
+ self.assertEqual(res['save_path'], '')
+ self.assertEqual(
+ (res['data'].all() == results_2[index]['data'].all()), True)
+ self.assertEqual(
+ (res['data'].all() == results_3[index]['data'].all()), True)
+
+
+if __name__ == '__main__':
+ main()
diff --git a/requirements.txt b/requirements.txt
index e068cdb3d9d4ad422153ac41a04eea22f052f989..5824da115cd595900900a4baba6d4bddb35f7744 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -4,7 +4,7 @@ yapf == 0.26.0
six >= 1.10.0
flask >= 1.1.0
flake8
-visualdl == 2.0.0a0
+visualdl >= 2.0.0b
cma >= 2.7.0
sentencepiece
colorlog