提交 8e05ffed 编写于 作者: D dyning

move out visulization from hubserving

上级 a5c095e0
...@@ -6,7 +6,6 @@ ...@@ -6,7 +6,6 @@
"use_gpu": true "use_gpu": true
}, },
"predict_args": { "predict_args": {
"visualization": false
} }
} }
}, },
......
...@@ -19,7 +19,7 @@ import numpy as np ...@@ -19,7 +19,7 @@ import numpy as np
import paddle.fluid as fluid import paddle.fluid as fluid
import paddlehub as hub import paddlehub as hub
from tools.infer.utility import draw_boxes, base64_to_cv2 from tools.infer.utility import base64_to_cv2
from tools.infer.predict_det import TextDetector from tools.infer.predict_det import TextDetector
...@@ -68,16 +68,12 @@ class OCRDet(hub.Module): ...@@ -68,16 +68,12 @@ class OCRDet(hub.Module):
def predict(self, def predict(self,
images=[], images=[],
paths=[], paths=[]):
draw_img_save='ocr_det_result',
visualization=False):
""" """
Get the text box in the predicted images. Get the text box in the predicted images.
Args: Args:
images (list(numpy.ndarray)): images data, shape of each is [H, W, C]. If images not paths 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 paths (list[str]): The paths of images. If paths not images
draw_img_save (str): The directory to store output images.
visualization (bool): Whether to save image or not.
Returns: Returns:
res (list): The result of text detection box and save path of images. res (list): The result of text detection box and save path of images.
""" """
...@@ -93,29 +89,21 @@ class OCRDet(hub.Module): ...@@ -93,29 +89,21 @@ class OCRDet(hub.Module):
all_results = [] all_results = []
for img in predicted_data: for img in predicted_data:
result = {'save_path': ''}
if img is None: if img is None:
logger.info("error in loading image") logger.info("error in loading image")
result['data'] = [] all_results.append([])
all_results.append(result)
continue continue
dt_boxes, elapse = self.text_detector(img) dt_boxes, elapse = self.text_detector(img)
print("Predict time : ", elapse) logger.info("Predict time : {}".format(elapse))
result['data'] = dt_boxes.astype(np.int).tolist()
if visualization: rec_res_final = []
image = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB)) for dno in range(len(dt_boxes)):
draw_img = draw_boxes(image, dt_boxes) rec_res_final.append(
draw_img = np.array(draw_img) {
if not os.path.exists(draw_img_save): 'text_region': dt_boxes[dno].astype(np.int).tolist()
os.makedirs(draw_img_save) }
saved_name = 'ndarray_{}.jpg'.format(time.time()) )
save_file_path = os.path.join(draw_img_save, saved_name) all_results.append(rec_res_final)
cv2.imwrite(save_file_path, draw_img[:, :, ::-1])
print("The visualized image saved in {}".format(save_file_path))
result['save_path'] = save_file_path
all_results.append(result)
return all_results return all_results
@serving @serving
...@@ -134,5 +122,5 @@ if __name__ == '__main__': ...@@ -134,5 +122,5 @@ if __name__ == '__main__':
'./doc/imgs/11.jpg', './doc/imgs/11.jpg',
'./doc/imgs/12.jpg', './doc/imgs/12.jpg',
] ]
res = ocr.predict(paths=image_path, visualization=True) res = ocr.predict(paths=image_path)
print(res) print(res)
\ No newline at end of file
...@@ -92,12 +92,24 @@ class OCRRec(hub.Module): ...@@ -92,12 +92,24 @@ class OCRRec(hub.Module):
if img is None: if img is None:
continue continue
img_list.append(img) img_list.append(img)
rec_res_final = []
try: try:
rec_res, predict_time = self.text_recognizer(img_list) rec_res, predict_time = self.text_recognizer(img_list)
for dno in range(len(rec_res)):
text, score = rec_res[dno]
rec_res_final.append(
{
'text': text,
'confidence': float(score),
}
)
except Exception as e: except Exception as e:
print(e) print(e)
return [] return [[]]
return rec_res
return [rec_res_final]
@serving @serving
def serving_method(self, images, **kwargs): def serving_method(self, images, **kwargs):
......
...@@ -6,7 +6,6 @@ ...@@ -6,7 +6,6 @@
"use_gpu": true "use_gpu": true
}, },
"predict_args": { "predict_args": {
"visualization": false
} }
} }
}, },
......
...@@ -19,7 +19,7 @@ import numpy as np ...@@ -19,7 +19,7 @@ import numpy as np
import paddle.fluid as fluid import paddle.fluid as fluid
import paddlehub as hub import paddlehub as hub
from tools.infer.utility import draw_ocr, base64_to_cv2 from tools.infer.utility import base64_to_cv2
from tools.infer.predict_system import TextSystem from tools.infer.predict_system import TextSystem
...@@ -68,18 +68,12 @@ class OCRSystem(hub.Module): ...@@ -68,18 +68,12 @@ class OCRSystem(hub.Module):
def predict(self, def predict(self,
images=[], images=[],
paths=[], paths=[]):
draw_img_save='ocr_result',
visualization=False,
text_thresh=0.5):
""" """
Get the chinese texts in the predicted images. Get the chinese texts in the predicted images.
Args: Args:
images (list(numpy.ndarray)): images data, shape of each is [H, W, C]. If images not paths 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 paths (list[str]): The paths of images. If paths not images
draw_img_save (str): The directory to store output images.
visualization (bool): Whether to save image or not.
text_thresh(float): the threshold of the recognize chinese texts' confidence
Returns: Returns:
res (list): The result of chinese texts and save path of images. res (list): The result of chinese texts and save path of images.
""" """
...@@ -93,53 +87,30 @@ class OCRSystem(hub.Module): ...@@ -93,53 +87,30 @@ class OCRSystem(hub.Module):
assert predicted_data != [], "There is not any image to be predicted. Please check the input data." assert predicted_data != [], "There is not any image to be predicted. Please check the input data."
cnt = 0
all_results = [] all_results = []
for img in predicted_data: for img in predicted_data:
result = {'save_path': ''}
if img is None: if img is None:
logger.info("error in loading image") logger.info("error in loading image")
result['data'] = [] all_results.append([])
all_results.append(result)
continue continue
starttime = time.time() starttime = time.time()
dt_boxes, rec_res = self.text_sys(img) dt_boxes, rec_res = self.text_sys(img)
elapse = time.time() - starttime elapse = time.time() - starttime
cnt += 1 logger.info("Predict time: {}".format(elapse))
print("Predict time of image %d: %.3fs" % (cnt, elapse))
dt_num = len(dt_boxes) dt_num = len(dt_boxes)
rec_res_final = [] rec_res_final = []
for dno in range(dt_num): for dno in range(dt_num):
text, score = rec_res[dno] text, score = rec_res[dno]
# if the recognized text confidence score is lower than text_thresh, then drop it rec_res_final.append(
if score >= text_thresh: {
# text_str = "%s, %.3f" % (text, score) 'text': text,
# print(text_str) 'confidence': float(score),
rec_res_final.append( 'text_region': dt_boxes[dno].astype(np.int).tolist()
{ }
'text': text, )
'confidence': float(score), all_results.append(rec_res_final)
'text_box_position': dt_boxes[dno].astype(np.int).tolist()
}
)
result['data'] = rec_res_final
if visualization:
image = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
boxes = dt_boxes
txts = [rec_res[i][0] for i in range(len(rec_res))]
scores = [rec_res[i][1] for i in range(len(rec_res))]
draw_img = draw_ocr(image, boxes, txts, scores, draw_txt=True, drop_score=0.5)
if not os.path.exists(draw_img_save):
os.makedirs(draw_img_save)
saved_name = 'ndarray_{}.jpg'.format(time.time())
save_file_path = os.path.join(draw_img_save, saved_name)
cv2.imwrite(save_file_path, draw_img[:, :, ::-1])
print("The visualized image saved in {}".format(save_file_path))
result['save_path'] = save_file_path
all_results.append(result)
return all_results return all_results
@serving @serving
...@@ -158,5 +129,5 @@ if __name__ == '__main__': ...@@ -158,5 +129,5 @@ if __name__ == '__main__':
'./doc/imgs/11.jpg', './doc/imgs/11.jpg',
'./doc/imgs/12.jpg', './doc/imgs/12.jpg',
] ]
res = ocr.predict(paths=image_path, visualization=False) res = ocr.predict(paths=image_path)
print(res) print(res)
\ No newline at end of file
...@@ -23,8 +23,14 @@ deploy/hubserving/ocr_system/ ...@@ -23,8 +23,14 @@ deploy/hubserving/ocr_system/
## 快速启动服务 ## 快速启动服务
以下步骤以检测+识别2阶段串联服务为例,如果只需要检测服务或识别服务,替换相应文件路径即可。 以下步骤以检测+识别2阶段串联服务为例,如果只需要检测服务或识别服务,替换相应文件路径即可。
### 1. 安装paddlehub ### 1. 准备环境
```pip3 install paddlehub --upgrade -i https://pypi.tuna.tsinghua.edu.cn/simple``` ```shell
# 安装paddlehub
pip3 install paddlehub --upgrade -i https://pypi.tuna.tsinghua.edu.cn/simple
# 设置环境变量
export PYTHONPATH=.
```
### 2. 安装服务模块 ### 2. 安装服务模块
PaddleOCR提供3种服务模块,根据需要安装所需模块。如: PaddleOCR提供3种服务模块,根据需要安装所需模块。如:
...@@ -75,7 +81,6 @@ $ hub serving start --modules [Module1==Version1, Module2==Version2, ...] \ ...@@ -75,7 +81,6 @@ $ hub serving start --modules [Module1==Version1, Module2==Version2, ...] \
"use_gpu": true "use_gpu": true
}, },
"predict_args": { "predict_args": {
"visualization": false
} }
} }
}, },
...@@ -99,32 +104,21 @@ hub serving start -c deploy/hubserving/ocr_system/config.json ...@@ -99,32 +104,21 @@ hub serving start -c deploy/hubserving/ocr_system/config.json
``` ```
## 发送预测请求 ## 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: 配置好服务端,可使用以下命令发送预测请求,获取预测结果:
```python ```python tools/test_hubserving.py server_url image_path```
import requests
import json
import cv2
import base64
def cv2_to_base64(image):
return base64.b64encode(image).decode('utf8')
# 发送HTTP请求
data = {'images':[cv2_to_base64(open("./doc/imgs/11.jpg", 'rb').read())]}
headers = {"Content-type": "application/json"}
# url = "http://127.0.0.1:8866/predict/ocr_det"
# url = "http://127.0.0.1:8866/predict/ocr_rec"
url = "http://127.0.0.1:8866/predict/ocr_system"
r = requests.post(url=url, headers=headers, data=json.dumps(data))
# 打印预测结果
print(r.json()["results"])
```
你可能需要根据实际情况修改`url`字符串中的端口号和服务模块名称。 需要给脚本传递2个参数:
- **server_url**:服务地址,格式为
`http://[ip_address]:[port]/predict/[module_name]`
例如,如果使用配置文件启动检测、识别、检测+识别2阶段服务,那么发送请求的url将分别是:
`http://127.0.0.1:8866/predict/ocr_det`
`http://127.0.0.1:8867/predict/ocr_rec`
`http://127.0.0.1:8868/predict/ocr_system`
- **image_path**:测试图像路径,可以是单张图片路径,也可以是图像集合目录路径
上面所示代码都已写入测试脚本,可直接运行命令:```python tools/test_hubserving.py``` 访问示例:
```python tools/test_hubserving.py http://127.0.0.1:8868/predict/ocr_system ./doc/imgs/```
## 自定义修改服务模块 ## 自定义修改服务模块
如果需要修改服务逻辑,你一般需要操作以下步骤(以修改`ocr_system`为例): 如果需要修改服务逻辑,你一般需要操作以下步骤(以修改`ocr_system`为例):
......
...@@ -117,16 +117,12 @@ def main(args): ...@@ -117,16 +117,12 @@ def main(args):
image_file_list = get_image_file_list(args.image_dir) image_file_list = get_image_file_list(args.image_dir)
text_sys = TextSystem(args) text_sys = TextSystem(args)
is_visualize = True is_visualize = True
tackle_img_num = 0
for image_file in image_file_list: for image_file in image_file_list:
img = cv2.imread(image_file) img = cv2.imread(image_file)
if img is None: if img is None:
logger.info("error in loading image:{}".format(image_file)) logger.info("error in loading image:{}".format(image_file))
continue continue
starttime = time.time() starttime = time.time()
tackle_img_num += 1
if not args.use_gpu and tackle_img_num % 30 == 0:
text_sys = TextSystem(args)
dt_boxes, rec_res = text_sys(img) dt_boxes, rec_res = text_sys(img)
elapse = time.time() - starttime elapse = time.time() - starttime
print("Predict time of %s: %.3fs" % (image_file, elapse)) print("Predict time of %s: %.3fs" % (image_file, elapse))
......
...@@ -91,7 +91,7 @@ def create_predictor(args, mode): ...@@ -91,7 +91,7 @@ def create_predictor(args, mode):
config.enable_use_gpu(args.gpu_mem, 0) config.enable_use_gpu(args.gpu_mem, 0)
else: else:
config.disable_gpu() config.disable_gpu()
config.enable_mkldnn() # config.enable_mkldnn()
config.set_cpu_math_library_num_threads(4) config.set_cpu_math_library_num_threads(4)
#config.enable_memory_optim() #config.enable_memory_optim()
config.disable_glog_info() config.disable_glog_info()
......
#!usr/bin/python # Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
# -*- coding: utf-8 -*- #
# 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
import sys
__dir__ = os.path.dirname(os.path.abspath(__file__))
sys.path.append(__dir__)
sys.path.append(os.path.abspath(os.path.join(__dir__, '..')))
from ppocr.utils.utility import initial_logger
logger = initial_logger()
import cv2
import numpy as np
import time
from PIL import Image
from ppocr.utils.utility import get_image_file_list
from tools.infer.utility import draw_ocr, draw_boxes
import requests import requests
import json import json
import cv2
import base64 import base64
import time
def cv2_to_base64(image): def cv2_to_base64(image):
return base64.b64encode(image).decode('utf8') return base64.b64encode(image).decode('utf8')
start = time.time()
# 发送HTTP请求 def draw_server_result(image_file, res):
data = {'images':[cv2_to_base64(open("./doc/imgs/11.jpg", 'rb').read())]} img = cv2.imread(image_file)
headers = {"Content-type": "application/json"} image = Image.fromarray(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
# url = "http://127.0.0.1:8866/predict/ocr_det" if len(res) == 0:
# url = "http://127.0.0.1:8866/predict/ocr_rec" return np.array(image)
url = "http://127.0.0.1:8866/predict/ocr_system" keys = res[0].keys()
r = requests.post(url=url, headers=headers, data=json.dumps(data)) if 'text_region' not in keys: # for ocr_rec, draw function is invalid
end = time.time() print("draw function is invalid for ocr_rec!")
return None
# 打印预测结果 elif 'text' not in keys: # for ocr_det
print(r.json()["results"]) print("draw text boxes only!")
print("time cost: ", end - start) boxes = []
for dno in range(len(res)):
boxes.append(res[dno]['text_region'])
boxes = np.array(boxes)
draw_img = draw_boxes(image, boxes)
return draw_img
else: # for ocr_system
print("draw boxes and texts!")
boxes = []
texts = []
scores = []
for dno in range(len(res)):
boxes.append(res[dno]['text_region'])
texts.append(res[dno]['text'])
scores.append(res[dno]['confidence'])
boxes = np.array(boxes)
scores = np.array(scores)
draw_img = draw_ocr(image, boxes, texts, scores, draw_txt=True, drop_score=0.5)
return draw_img
def main(url, image_path):
image_file_list = get_image_file_list(image_path)
is_visualize = False
headers = {"Content-type": "application/json"}
cnt = 0
total_time = 0
for image_file in image_file_list:
img = open(image_file, 'rb').read()
if img is None:
logger.info("error in loading image:{}".format(image_file))
continue
# 发送HTTP请求
starttime = time.time()
data = {'images':[cv2_to_base64(img)]}
r = requests.post(url=url, headers=headers, data=json.dumps(data))
elapse = time.time() - starttime
total_time += elapse
print("Predict time of %s: %.3fs" % (image_file, elapse))
res = r.json()["results"][0]
# print(res)
if is_visualize:
draw_img = draw_server_result(image_file, res)
if draw_img is not None:
draw_img_save = "./server_results/"
if not os.path.exists(draw_img_save):
os.makedirs(draw_img_save)
cv2.imwrite(
os.path.join(draw_img_save, os.path.basename(image_file)),
draw_img[:, :, ::-1])
print("The visualized image saved in {}".format(
os.path.join(draw_img_save, os.path.basename(image_file))))
cnt += 1
if cnt % 100 == 0:
print(cnt, "processed")
print("avg time cost: ", float(total_time)/cnt)
if __name__ == '__main__':
if len(sys.argv) != 3:
print("Usage: %s server_url image_path" % sys.argv[0])
else:
server_url = sys.argv[1]
image_path = sys.argv[2]
main(server_url, image_path)
\ No newline at end of file
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册