def dealPicture(picSrc):
print("picSrc:%s"%picSrc)
result_dir = './test_result'
if os.path.exists(result_dir):
shutil.rmtree(result_dir)
os.mkdir(result_dir)
resultString = ""
image_file = Uploadpic + picSrc
print("image_file:%s"%image_file)
image = np.array(Image.open(image_file).convert('RGB'))
t = time.time()
//利用原有模型实现识别
result, image_framed = ocr.model(image)
//利用pytesseract实现中文识别
#result, image_framed = ocr.model2(image)
output_file = os.path.join(result_dir, image_file.split('/')[-1])
Image.fromarray(image_framed).save(output_file)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("\nRecognition Result:\n")
for key in result:
print(result[key][1])
resultString += result[key][1]
resultString = re.sub("[#$%&\'()*+,-./:;<=>?@，。?★、…【】《》？“”‘’！[\\]^_`{|}~!]+", "",resultString)
print("resultString:%s"%resultString)
return resultString
def post(self, *args, **kwargs):
ret = {'result': 'OK'}
result_dir = './test_result'
if os.path.exists(result_dir):
shutil.rmtree(result_dir)
os.mkdir(result_dir)
upload_path = os.path.join(os.path.dirname(__file__),
'demo') # 文件的暂存路径
file_metas = self.request.files.get('file',
None) # 提取表单中‘name’为‘file’的文件元数据
filename = file_metas[0]['filename']
file_path = os.path.join(upload_path, filename)
with open(file_path, 'wb') as up:
up.write(file_metas[0]['body'])
image = np.array(Image.open(file_path).convert('RGB'))
t = time.time()
result, image_framed = ocr.model(image)
output_file = os.path.join(result_dir, file_path.split('/')[-1])
Image.fromarray(image_framed).save(output_file)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("单条预测时间", time.time() - t)
print("\nRecognition Result:\n")
for key in result:
print(result[key][1])
self.write("13213")
def OCRconsumer(q, targetPath):
result_str = ""
while True:
# if q.empty():
# print(1)
# continue
res = q.get()
if res is None: break
tic = time.time()
id, image = res
result, image_framed = ocr.model(image)
toc = time.time()
print("one pic predict complete, it took {:.3f}s".format(toc - tic))
j = 0
result_str += str(id)
result_str += '\n'
for key in result:
if j == 0 or j == 1:
str_tmp = str(result[key][1])
result_str += str_tmp
result_str += '\n'
else:
break
j += 1
result_str += "###\n"
txt_path = targetPath + "/OnlineShopLicense.txt"
writer = open(txt_path, 'a')
writer.write(result_str)
def predict(file):
image = cv2.imread(file)
t = time.time()
result, image_framed = ocr.model(image)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("\n Recognition Result: \n")
for key in result:
print(result[key][1])
return result[key][1]
def ocr_img_pse():
file = request.files['file']
lan = request.args.get('language', 'CHE')
angle = request.args.get('angle', 'True')
logging.info('语言类型%s' % lan)
logging.info('是否需要角度%s' % angle)
if angle == 'False':
angle = False
else:
angle = True
if not file:
logging.info('图片为空')
return response(json.dumps({'code': -1, 'msg': '文件为空', 'result': ''}))
#if file.filename.split('.')[-1] not in ['jpg', 'png', 'jpeg']:
# logging.info('图片格式错误：%s' % file.filename)
# return response(json.dumps({'code': -2, 'msg': '文件格式错误', 'result': ''}))
logging.info('图片格式：%s' % file.filename)
# save_path = 'up_images/' + file.filename
# file.save('mid.jpg')
# img = cv2.imread(save_path, 3)
try:
img_buffer = np.asarray(bytearray(file.read()), dtype='uint8')
bytesio = BytesIO(img_buffer)
img = skimage.io.imread(bytesio)
except Exception as e:
logging.info(str(e), exc_info=True)
return response(json.dumps({
'code': -2,
'msg': '文件格式错误',
'result': ''
}))
if len(img.shape) == 2:
img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
elif img.shape[-1] == 4:
img = cv2.cvtColor(img, cv2.COLOR_BGRA2BGR)
else:
pass
a = time.time()
# cv2.imwrite('mid'+str(a)+'.jpg',img)
results, img_shape = model(img, lan, angle)
# logging.info('图片处理完成:'+str(results))
# return json.dumps({'code':0, 'msg':'', 'result':results, 'shape':img_shape})
return response(
json.dumps({
'code': 0,
'msg': '',
'result': results,
'shape': img_shape
}))
def imgArrOCR(image):
'''
do ocr operations on a single image numpy array
return a single string of ocr result
lines split by '\\n'
'''
result, image_framed = ocr.model(image)
result_str = ""
for key in result:
result_str += result[key][1]
result_str += '\n'
return result_str
def detect(image):
t = time.time()
result, image_framed = ocr.model(image)
# # save output file for debug
# output_file = os.path.join(result_dir, os.path.split(image_file)[-1])
# Image.fromarray(image_framed).save(output_file)
# print debug message
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("Recognition Result:")
for rect, text in result:
print(text)
return result
def rec_img(task_id):
if task_id == "file":
img_file = request.form['image_path']
if os.path.exists(img_file):
image = np.array(Image.open(img_file).convert('RGB'))
t = time.time()
result, image_framed = ocr.model(image)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
re_dict = {}
for key in result:
re_dict[result[key][1]] = list(result[key][0])
return json.dumps(re_dict)
else:
return 'Image does not exist!'
def otr():
if request.method == 'POST':
f = request.files['file']
if not (f and allowed_file(f.filename)):
return jsonify({
"error": 1001,
"msg": "请检查上传的图片类型，仅限于png、PNG、jpg、JPG、bmp"
})
user_input = request.form.get("name")
# 当前文件所在路径
basepath = os.path.dirname(__file__)
# 注意：没有的文件夹一定要先创建，不然会提示没有该路径
upload_path = os.path.join(basepath, 'static/images',
secure_filename(f.filename))
f.save(upload_path)
# --------------------------------------------------------------------------
# image = np.array(Image.open(upload_path).convert('RGB'))
image = cv2.imread(upload_path)
t = time.time()
ot_image = antiMissingMain.getOTDetectROI(image)
result, image_framed = ocr.model(ot_image)
output_file = os.path.join("test_result", upload_path.split('/')[-1])
Image.fromarray(image_framed).save(output_file)
Image.fromarray(image_framed).save("static/images/test.jpg")
stringResult = {}
stringResult[0] = "Mission complete, it took {:.3f}s".format(
time.time() - t)
i = 1
for key in result:
stringResult[i] = result[key][1]
i = i + 1
# os.remove(upload_path) # 删除已识别的图片
# --------------------------------------------------------------------------
image_framed = cv2.cvtColor(np.asarray(image_framed),
cv2.COLOR_RGB2BGR)
cv2.imwrite(os.path.join(basepath, 'static/images', 'test.jpg'),
image_framed)
return render_template('otr_ok.html',
userinput=user_input,
val1=time.time(),
data_dict=stringResult)
return render_template('otr.html')
def pic2word(pic):
# image = np.array(Image.open(image_file).convert('RGB'))
image = np.array(pic)
t = time.time()
# result: the text
# image_framed: the image with segmentation
result, image_framed = ocr.model(image)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("\nRecognition Result:\n")
f1 = open('word.txt', 'w')
f2 = open('to_english.txt', 'w')
for key in result:
# print(result[key][1])
f2.write(result[key][1])
f1.write(result[key][1] + '\n')
def fileOCR(imagePath):
'''
do ocr operations on a single image file
return a single string of ocr result
lines split by '\\n'
'''
image = np.array(PIL.Image.open(imagePath).convert('RGB'))
# print(image.shape())
result, image_framed = ocr.model(image)
result_str = ""
for key in result:
result_str += result[key][1]
result_str += '\n'
return result_str
def predict(path):
t = time.time()
image = np.array(Image.open(path).convert('RGB'))
result, _ = ocr.model(image)
result = ' '.join([result[key][1] for key in result])
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("Recognition Result:", result)
response = make_response(result)
response.headers['Access-Control-Allow-Origin'] = '*'
response.headers['Access-Control-Allow-Methods'] = 'POST'
response.headers[
'Access-Control-Allow-Headers'] = 'x-requested-with,content-type'
return response
def ocr_result(arg):
image = np.array(
Image.open('/'.join(['./image_input', arg])).convert('RGB'))
t = time.time()
result, image_framed = ocr.model(image)
output_file = '/'.join(['./image_result', arg + '.txt'])
# Image.fromarray(image_framed).save(output_file)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("\nRecognition Result:\n")
txt = ''
for key in result:
txt = txt + result[key][1]
print(txt)
with open(output_file, 'wb') as f:
f.write(txt)
f.close()
return txt
def identify_text():
if request.method == 'POST':
f = request.files['file']
# basepath = os.path.dirname(__file__) # 当前文件所在路径
# upload_path = os.path.join(basepath, 'static/uploads',f.filename) # 注意：没有的文件夹一定要先创建，不然会提示没有该路径
upload_path = os.path.join(os.getcwd(), 'static/uploads/', f.filename)
print(upload_path)
f.save(upload_path)
image = cv2.imread(upload_path, 1)
#result=ocrText(upload_path)
res=''
result, image_framed = ocr.model(image)
for key in result:
print(result[key][1])
res = res+(result[key][1])
if (os.path.exists(upload_path)):
os.remove(upload_path)
return jsonify({'result': res})
def demo(image, image_name='initial_name.jpg'):
psenet_start_time = time.time()
psenet_text_recs, psenet_draw_image = psenet_predict(image)
psenet_end_time = time.time()
print("psenet耗时：|{}|".format(psenet_end_time - psenet_start_time))
psenet_image_name = image_name[:-4] + '_psenet' + ".jpg"
print("psenet_image_name:|{}|,psenet_result_text_recs:|{}|".format(
psenet_image_name, psenet_text_recs))
objectid = container_type(text_recs=psenet_text_recs)
psenet_result_path = os.path.join(psenet_result_dir, psenet_image_name)
#@@@@@@
Image.fromarray(psenet_draw_image).save(psenet_result_path) #处理十万图片不输出
# ==================
# deeptext 文字识别 #deeptext的image传入ndarray类型，但在识别的时候需要PIL类型
# ===================
adjust_result_dir = os.path.join(adjust_main_dir, psenet_image_name[:-4])
makedir(adjust_result_dir)
ocr_result = []
ocr_result, cfd_scores = ocr.model(
image=image,
text_recs=psenet_text_recs,
adjust_result_dir=adjust_result_dir,
original_image_name=psenet_image_name,
)
print("psenet_image_name:|{}|,ocr_result:|{}|".format(
psenet_image_name, ocr_result))
hscore_text_recs = []
hscore_ocr_result = []
# print("====cfd_score:|{}|".format( cfd_scores ))
for index, confidence_score in enumerate(cfd_scores):
# print( "=======confidence:|{}|".format( confidence_score))
if confidence_score >= 0.1:
hscore_text_recs.append(psenet_text_recs[index])
hscore_ocr_result.append(ocr_result[index])
else:
print("confidence_score:|{}| < 0.1 mv".format(confidence_score))
ocr_result_file.writelines(psenet_image_name + ' ' + str(ocr_result) +
',' + '\n')
psenet_point_file.writelines(psenet_image_name + ':' +
str(psenet_text_recs) + '\n')
return hscore_text_recs, hscore_ocr_result, objectid
def get(self):
start = time.time()
result_dir = './test_result'
if os.path.exists(result_dir):
shutil.rmtree(result_dir)
os.mkdir(result_dir)
for image_file in sorted(image_files):
image = np.array(Image.open(image_file).convert('RGB'))
t = time.time()
result, image_framed = ocr.model(image)
output_file = os.path.join(result_dir, image_file.split('/')[-1])
Image.fromarray(image_framed).save(output_file)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("单条预测时间", time.time() - t)
print("\nRecognition Result:\n")
for key in result:
print(result[key][1])
print("总计预测时间", time.time() - start)
self.write("1233")
def get(self):
args = self.reqparse.parse_args()
image_url = args['image_url']
image_file = glob(image_url)
if len(image_file) < 1:
print("")
return Output.error("文件不存在！请先上传文件再进行识别！")
image = np.array(Image.open(image_file).convert('RGB'))
t = time.time()
result, image_framed = ocr.model(image)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("\nRecognition Result:\n")
data = ''
for key in result:
print(result[key][1])
# data.append(result[key][1])
data += result[key][1]
# val.val()
return Output.success(data)
def predict():
for image_file in sorted(image_files):
image = np.array(Image.open(image_file).convert('RGB'))
result, image_framed = ocr.model(image)
output_file = os.path.join(result_dir, image_file.split('/')[-1])
Image.fromarray(image_framed).save(output_file)
for key in result:
write_path = os.getcwd()+'/test_result/result.txt'
result_name = image_file.split('/')[-1]
result_name = result_name.split('.')[0]
print(result_name+':'+result[key][1])
f1 = open(write_path,'a')
f1.write(result_name+':'+result[key][1])
f1.write('\n')
def dirOCR(dirpath):
'''
do ocr operation on specific directory
return a list contains tuples which likes (filename,ocrresultstring)
'''
filelist = os.listdir(dirpath)
results = []
for filename in filelist:
if os.path.isfile(filename):
imagePath = os.path.join(dirpath, filename)
try:
image = np.array(PIL.Image.open(imagePath).convert('RGB'))
result, image_framed = ocr.model(image)
result_str = ""
for key in result:
result_str += result[key][1]
result_str += '\n'
results.append((filename, result_str))
except Exception as e:
print(e)
continue
return results
def run(image_file, type):
# if len(sys.argv)==3:
# image_file=sys.argv[1]
# type=sys.argv[2]
# else:
# print('wrong!!!')
image = np.array(Image.open(image_file).convert('RGB'))
t = time.time()
result, image_framed = ocr.model(image)
print(t)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("\nRecognition Result:\n")
result_string = ''
for key in result:
result_string += result[key][1]
dict = {
'certificate_of_degree': deal_certificate_of_degree,
'patent': deal_patent,
'IDcard': deal_IDcard,
'business_license': deal_business_license
}
result_data = dict[type](result_string)
return result_data
def post(self):
files = self.request.files.get('file', None)
lan = self.get_argument('language', 'CHE')
angle = self.get_argument('angle', 'True')
combine = self.get_argument('combine', 'False')
lines = self.get_argument('lines', [])
just_detection = self.get_argument('just_detection','False')
if lines:
lines = eval(lines)
if not isinstance(lines, list):
lines = []
if not files:
logging.info('图片为空')
self.write(self.resp({'code':-1, 'msg': '文件为空', 'result': ''}))
self.finish()
file = files[0]
logging.info('表格线：%s' % str(lines))
logging.info('文件%s' % file.filename)
logging.info('语言类型%s' % lan)
logging.info('是否需要角度%s' % angle)
logging.info('是否需要连接%s' % combine)
logging.info('只做检测%s' % just_detection)
if angle == 'False':
angle = False
else:
angle = True
if combine == 'False':
combine = False
else:
combine = True
just_detection = False if just_detection == 'False' else True
try:
img_buffer = np.asarray(bytearray(file.body), dtype='uint8')
bytesio = BytesIO(img_buffer)
img = skimage.io.imread(bytesio)
logging.info('io图片完成!')
except Exception as e:
logging.info(str(e), exc_info=True)
self.write(self.resp({'code': -2, 'msg': '文件格式错误', 'result': ''}))
self.finish()
if len(img.shape) == 2:
img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
elif img.shape[-1] == 4:
img = cv2.cvtColor(img, cv2.COLOR_BGRA2BGR)
else:
pass
try:
start = time.time()
results, img_shape = model(img, lan, angle, combine, lines ,just_detection)
end = time.time()
logging.info('ocr total time %s' % str(end-start))
#logging.info(results)
self.write(self.resp({'code':0, 'msg': '', 'result': results, 'shape': img_shape}))
except:
logging.info('error:{}'.format(straceback.format_exc()), exc_info=True)
self.write(self.resp({'code': -3, 'msg': '模型出错', 'result': ''}))
self.finish()
# + getLine(img, (Point[0] + 1, Point[1] + 1))
# else:
# return ()
if __name__ == '__main__':
result_dir = './test_result'
if os.path.exists(result_dir):
shutil.rmtree(result_dir)
os.mkdir(result_dir)
for image_file in sorted(image_files):
AllLine = []
image = np.array(Image.open(image_file).convert('RGB'))
t = time.time()
result, image_framed, boxes, scale = ocr.model(image)
output_file = os.path.join(result_dir, image_file.split('/')[-1])
Image.fromarray(image_framed).save(output_file)
img = cv2.resize(image,
None,
None,
fx=scale,
fy=scale,
interpolation=cv2.INTER_LINEAR)
img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# img = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("\nRecognition Result:\n")
for box in boxes:
for i in range(box[1], box[5]):
difference = cv2.subtract(im, im2)
result = not np.any(difference)
while (not result):
x, y = im.shape[0:2]
im = im[0:lastrose - 22, ]
# plt.figure(figsize=(20, 20))
# plt.imshow(im)
# plt.show()
print(im.shape)
im = cv2.resize(im, (int(im.shape[1] * 2), int(im.shape[0] * 2)))
for j in range(len(sortrose)):
sortrose[j] = sortrose[j] * 2
cv2.bitwise_not(im, im)
shape = im.shape
print(shape)
result, image_framed, isname = ocr.model(im, sortrose, nameMap)
print("\nRecognition Result:\n")
total = []
str1 = ""
flag = False
for key in result:
out = result[key][1]
# print(str(isname[key]) + ":" + out)
if isname[key] == 1:
if str1 != "":
if str1.find(":") != -1:
if str1.split(":")[-1] != "":
total.append(str1)
str1 = ""
else:
if str1 != "":
def bank_flow_identity(imagePath): # imagePath './test_images/gs.jpg'
image = cv2.imread(imagePath, 1)
print(image)
#二值化
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
binary = cv2.adaptiveThreshold(~gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
cv2.THRESH_BINARY, 35, -5)
rows, cols = binary.shape
scale = 40
#识别横线
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (cols // scale, 1))
eroded = cv2.erode(binary, kernel, iterations=1)
dilatedcol = cv2.dilate(eroded, kernel, iterations=1)
#识别竖线
scale = 20
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (1, rows // scale))
eroded = cv2.erode(binary, kernel, iterations=1)
dilatedrow = cv2.dilate(eroded, kernel, iterations=1)
#标识交点
bitwiseAnd = cv2.bitwise_and(dilatedcol, dilatedrow)
#标识表格
merge = cv2.add(dilatedcol, dilatedrow)
#识别黑白图中的白色点
ys, xs = np.where(bitwiseAnd > 0)
mylisty = []
mylistx = []
#通过排序，获取跳变的x和y的值，说明是交点，否则交点会有好多像素值，我只取最后一点
i = 0
myxs = np.sort(xs)
for i in range(len(myxs) - 1):
if (myxs[i + 1] - myxs[i] > 10):
mylistx.append(myxs[i])
i = i + 1
mylistx.append(myxs[i])
# print(mylistx)
# print(len(mylistx))
i = 0
myys = np.sort(ys)
#print(np.sort(ys))
for i in range(len(myys) - 1):
if (myys[i + 1] - myys[i] > 20):
mylisty.append(myys[i])
i = i + 1
mylisty.append(myys[i])
i = 0
t = time.time()
res = ''
for i in range(13): #只有13行有效数字
ROI = image[mylisty[i]:mylisty[i + 1], mylistx[9]:mylistx[10]]
result, image_framed = ocr.model(ROI)
for key in result:
print(result[key][1])
res = res + result[key][1] + '||'
i = i + 1
print("tsk complete, it took {:.3f}s".format(time.time() - t))
return res
def ocr_img_pse():
print('request.files.keys()',list(request.files.keys()))
lan = request.args.get('language')
print('test_lan',lan)
#lan = 'jpe'
#lan = 'eng'
#lan = 'chn'
#print('test_lan',lan)
logging.info('语言类型%s' % lan)
if 'json_response' in request.files.to_dict().keys():
json_response = request.files['json_response']
print('yes')
if json_response:
print('aaaaa')
elif 'file' in request.files.to_dict().keys():
try:
lan = request.form['language']
print('lan',request.form['language'])
except:
lan = request.args.get('language')
print('test_lan',lan)
file = request.files['file']
#llann = request.files['language']
#print('language',llann)
if not file:
logging.info('图片为空')
return response(json.dumps({'code': -1, 'msg': '文件为空', 'result': ''}))
logging.info('图片格式：%s' % file.filename)
try:
img_buffer = np.asarray(bytearray(file.read()), dtype='uint8')
bytesio = BytesIO(img_buffer)
img = skimage.io.imread(bytesio)
except Exception as e:
logging.info(str(e), exc_info=True)
return response(json.dumps({'code': -2, 'msg': '文件格式错误', 'result': ''}))
if len(img.shape) == 2:
img = cv2.cvtColor(img, cv2.COLOR_GRAY2BGR)
elif img.shape[-1] == 4:
img = cv2.cvtColor(img, cv2.COLOR_BGRA2BGR)
else:
pass
a = time.time()
try:
check_label = request.form['checkLabel']
print('checkLabel', check_label)
except:
check_label = 'False'
#check_label = False
file_url = 'http://39.104.88.168/image_rec/'
if check_label == 'False':
if OCR:
results,img_shape, after_detect_img_name = model(img, lan,file.filename)
return response(json.dumps({'code':0, 'msg':'','file_url':file_url,'img_info':results,'after_detect_img_info':after_detect_img_name},cls=MyEncoder))
else:
recs, img_shape, after_detect_img_name,rec_name = model(img, lan, file.filename)
return response(json.dumps({'code': 0, 'msg': '', 'file_url': file_url, 'img_info': recs,
'after_detect_img_info': after_detect_img_name, 'rec_name': rec_name},
cls=MyEncoder))
else:
after_detect_img_name = file.filename
results = []
txt_file = request.form['txtlabel']
print(txt_file)
if txt_file == 'True' :
check_label_text = open('check_label/label.txt','r').readlines()
#random.shuffle(check_label_text)
check_label_text = check_label_text
for check_label_line in check_label_text:
label_img_info = {}
label_img_info['img_name'] = check_label_line.split(' ')[0]+'.jpg'
label_and_rec_text = {}
#label_and_rec_text['label']= check_label_line['label_text']
label_and_rec_text['label']= ' '.join(check_label_line.split(' ')[1:]).strip()
label_and_rec_text['rec_text'] = '1'
label_img_info['text'] = label_and_rec_text
results.append(label_img_info)
else:
check_json_file = request.form['filename']
check_label_text = open(check_json_file +'.json','r').readlines()
# random.shuffle(check_label_text)
for check_label_line in check_label_text:
check_label_line = json.loads(check_label_line)
label_img_info = {}
label_img_info['img_name'] = check_label_line['img_name']
label_and_rec_text = {}
label_and_rec_text['label']= check_label_line['text']['label']
label_and_rec_text['rec_text']= check_label_line['text']['rec_text']
label_img_info['text'] = label_and_rec_text
results.append(label_img_info)
return response(json.dumps({'code':0, 'msg':'','file_url':file_url,'img_info':results,'after_detect_img_info':after_detect_img_name}))
else:
label_data = request.data
print('label_data',label_data)
if label_data != '':
label_data = json.loads(label_data)
aaaa =label_data['text']
json_label_path = get_json_path(lan)
print('.'.join(label_data['imgurl'].split('/')[-1].split('.')[:-1]))
jsonname = json_label_path + '.'.join(label_data['imgurl'].split('/')[-1].split('.')[:-1]) + '.json'
with open(jsonname,'w') as f:
json.dump(label_data,f)
print('保存成功')
return response(json.dumps({'code':-1,'msg':'保存成功'}))
#coors_files = glob('./test_coors/*.*')
def crop(image_path, coors, save_dir):
image_name = os.path.split(image_path)[1].split('.')[0]
image = cv2.imread(image_path)
for index, coor in enumerate(coors):
x1, y1, x2, y2 = coor[0], coor[1], coor[6], coor[7]
#print(x1, y1, x2, y2)
roi_image = image[y1: y2 + 1, x1: x2 + 1]
cv2.imwrite(os.path.join(save_dir, image_name+'_{}.jpg'.format(index)), roi_image)
if __name__ == '__main__':
result_dir = './test_result'
if os.path.exists(result_dir):
shutil.rmtree(result_dir)
os.mkdir(result_dir)
for image_file in sorted(image_files):
image = np.array(Image.open(image_file).convert('RGB'))
t = time.time()
result, image_framed , text_recs = ocr.model(image)
output_file = os.path.join(result_dir, image_file.split('/')[-1])
Image.fromarray(image_framed).save(output_file)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("\nRecognition Result:\n")
for key in result:
print(result[key][1])
# crop(image_file,text_recs,'./test_coors')
def up_image():
# curl -v http://127.0.0.1:9436/ocr -F "image_file=@./test.jpg"
if request.method == 'POST' and request.files.get('image_file'):
timec = str(time.time()).replace(".", "")
file = request.files.get('image_file')
print(file)
img_data = file.read()
img = BytesIO(img_data)
img = Image.open(img, mode="r")
# username = request.form.get("name")
size = img.size
print("接收图片尺寸: {}".format(size))
if size[0] < 20 or size[1] < 14:
content = json.dumps({
"error_code": 1003,
'error_message': 'file to small'
})
resp = response_headers(content)
return resp
# 保存图片
file_name = "{}_{}.{}".format('temp', timec, 'jpg')
print("保存图片： {}".format(file_name))
file_path = os.path.join(api_image_dir + file_name)
with open(file_path, 'wb') as f:
f.write(img_data)
f.close()
s = time.time()
value = []
try:
image = np.array(img.convert('RGB'))
t = time.time()
result, image_framed, scale = ocr.model(image)
file_name = "{}_{}.{}".format('ocr_', timec, 'jpg')
dest_path = os.path.join(api_done_dir + file_name)
# os.system('mv '+file_path+' '+dest_path)
Image.fromarray(image_framed).save(dest_path)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("\nRecognition Result:\n")
texts = []
for key in result:
print(result[key][1])
texts.append(result[key][1])
value.append({
'loc': [int(x) for x in result[key][0]],
'rate': result[key][2],
'text': result[key][1]
})
e = time.time()
traceback.print_exc()
print("识别结果: {}".format(json.dumps(value, ensure_ascii=False)))
result = {
'error_code': 0,
'time': timec, # 时间戳
'value': value, # 预测的结果
'text': u'\n'.join(texts),
'scale': scale, #
'author': 'sloanyyc',
'speed_time(ms)': int((e - s) * 1000) # 识别耗费的时间
}
return json.dumps(result, ensure_ascii=False)
except Exception as ex:
e = time.time()
print('识别错误', ex)
result = {
'error_code': 1004,
'time': timec, # 时间戳
'value': [], # 预测的结果
'scale': 1, #
'author': 'sloanyyc',
'speed_time(ms)': int((e - s) * 1000) # 识别耗费的时间
}
return json.dumps(result, ensure_ascii=False)
else:
content = json.dumps({
"error_code": 1001,
'error_message': 'only file via form post support'
})
resp = response_headers(content)
return resp
'author': 'sloanyyc',
'speed_time(ms)': int((e - s) * 1000) # 识别耗费的时间
}
return json.dumps(result, ensure_ascii=False)
except Exception as ex:
e = time.time()
print('识别错误', ex)
result = {
'error_code': 1004,
'time': timec, # 时间戳
'value': [], # 预测的结果
'scale': 1, #
'author': 'sloanyyc',
'speed_time(ms)': int((e - s) * 1000) # 识别耗费的时间
}
return json.dumps(result, ensure_ascii=False)
else:
content = json.dumps({
"error_code": 1001,
'error_message': 'only file via form post support'
})
resp = response_headers(content)
return resp
if __name__ == '__main__':
first_load_image = Image.open('test.png', mode="r")
ocr.model(np.array(first_load_image.convert('RGB')))
app.run(host='0.0.0.0', debug=False, port=9436)
#-*- coding:utf-8 -*-
import os
import ocr
import time
import shutil
import numpy as np
from PIL import Image
from glob import glob
image_files = glob('./test_images/*.*')
if __name__ == '__main__':
result_dir = './test_result'
if os.path.exists(result_dir):
shutil.rmtree(result_dir)
os.mkdir(result_dir)
for image_file in sorted(image_files):
image = np.array(Image.open(image_file).convert('RGB'))
t = time.time()
result, image_framed = ocr.model(image)
output_file = os.path.join(result_dir, image_file.split('/')[-1])
Image.fromarray(image_framed).save(output_file)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("\nRecognition Result:\n")
import pdb
pdb.set_trace()
for key in result:
print(result[key][1])
import os
import codecs
import sys
if __name__ == '__main__':
# TODO: you should change the path
if sys.argv[1] == 'test':
name = 'test'
else:
name = 'train'
image_files = glob('../../data/' + name + '_images/*.*')
with codecs.open('../../data/result_' + name + '_ocr.txt', 'w', encoding='utf-8') as f:
t = time.time()
for image_file in sorted(image_files):
filename = os.path.split(image_file)[1]
image = np.array(Image.open(image_file).convert('RGB'))
result, image_framed = ocr.model(image)
text = filename
for key in result:
text += '\t'
text += result[key][1]
text += '\n'
print(text)
f.write(text)
print('------> total cost', time.time() - t, '------->images', len(image_files))
print('done')
#-*- coding:utf-8 -*-
import os
import ocr
import time
import shutil
import numpy as np
from PIL import Image
from glob import glob
image_files = glob('./test_images/*.*')
if __name__ == '__main__':
result_dir = './test_result'
# if os.path.exists(result_dir):
# shutil.rmtree(result_dir)
# os.mkdir(result_dir)
for image_file in sorted(image_files):
image = np.array(Image.open(image_file).convert('RGB'))
t = time.time()
result, image_framed = ocr.model(image, detectAngle=True)
# output_file = os.path.join(result_dir, image_file.split('/')[-1])
# Image.fromarray(image_framed).save(output_file)
print("Mission complete, it took {:.3f}s".format(time.time() - t))
print("\nRecognition Result:\n")
for key in result:
print(result[key][1])
