def train_ticket(filepath):
scale, maxScale = IMGSIZE[0], 2048
from run.text.keras_detect import text_detect
from run.main import TextOcrModel
angle_detect = None
model = TextOcrModel(ocr, text_detect, angle_detect)
img = cv.imread(filepath)
result, angle = model.model(
img,
scale=scale,
maxScale=maxScale,
detectAngle=False, ##是否进行文字方向检测
MAX_HORIZONTAL_GAP=100, ##字符之间的最大间隔
MIN_V_OVERLAPS=0.6,
MIN_SIZE_SIM=0.6,
TEXT_PROPOSALS_MIN_SCORE=0.1,
TEXT_PROPOSALS_NMS_THRESH=0.3,
TEXT_LINE_NMS_THRESH=0.99, ##iou值
LINE_MIN_SCORE=0.1,
leftAdjustAlph=0.01, ##
rightAdjustAlph=0.01, ##
)
res = trainTicket.trainTicket(result)
res = res.res
res = [{'text': res[key], 'name': key, 'box': {}} for key in res]
return res
nclass,
256,
leakyRelu=False,
lstmFlag=LSTMFLAG,
GPU=GPU,
alphabet=alphabet)
if os.path.exists(ocrModel):
crnn.load_weights(ocrModel)
else:
print("download model or tranform model with tools!")
ocr = crnn.predict_job
from main import TextOcrModel
model = TextOcrModel(ocr, text_detect, angle_detect)
billList = ['通用OCR', '火车票', '身份证']
class OCR:
"""通用OCR识别"""
def GET(self):
post = {}
post['postName'] = 'ocr' ##请求地址
post['height'] = 1000
post['H'] = 1000
post['width'] = 600
post['W'] = 600
post['billList'] = billList
return render.ocr(post)
alphabet=alphabet)
print('[INFO] Successfully initialize CRNN recognizer')
if os.path.exists(ocrModel):
crnn.load_weights(ocrModel)
print("[INFO] Successfully load Torch-ocr model...")
else:
print("download model or tranform model with tools!")
ocr = crnn.predict_job
if __name__ == "__main__":
from main import TextOcrModel
# initialize model pipeline
model = TextOcrModel(ocr, text_detect, angle_detect)
# arguments
textAngle = False
# load image
img = cv2.imread(args["image_file"])
# predict
detectAngle = textAngle
result, angle = model.model(
img,
scale=scale,
maxScale=maxScale,
detectAngle=detectAngle, ##是否进行文字方向检测，通过web传参控制
MAX_HORIZONTAL_GAP=100, ##字符之间的最大间隔，用于文本行的合并
MIN_V_OVERLAPS=0.6,
MIN_SIZE_SIM=0.6,
def currency_invoice(img, arg, arg2, model_flg):
from main import TextOcrModel
angle_detect = None
model = TextOcrModel(ocr, text_detect, angle_detect)
# print(arg[12],'---------------------',type(arg[12]))
if model_flg == True:
model = TextOcrModel(ocr, text_detect, angle_detect)
rotate_img = color_filter(img, color_thre=255, mode='less')
img = color_filter(rotate_img, color_thre=50, mode='more')
if lab == '-1':
result, angle, im_show = model.model_CRAFT(
img,
scale=arg[0],
detectAngle=False, ##是否进行文字方向检测
MAX_HORIZONTAL_GAP=arg[2], ##字符之间的最大间隔 #30
MIN_V_OVERLAPS=arg[3], #0.5
TEXT_PROPOSALS_MIN_SCORE=arg[4], #[0.5, 0.2, 0.7],
leftAdjustAlph=arg[6], ##
rightAdjustAlph=arg[7], ##
Adjustbox=arg[5], ##
pixel_filter=arg[8], ##
batch_by_1=arg[9],
scoremap_enhance_pixel=arg[10],
)
elif lab == '0':
result, angle, im_show = model.model_CRAFT(
img,
scale=arg[0],
detectAngle=False, ##是否进行文字方向检测
MAX_HORIZONTAL_GAP=arg[2], ##字符之间的最大间隔 #30
MIN_V_OVERLAPS=arg[3], #0.5
TEXT_PROPOSALS_MIN_SCORE=arg[4], #[0.5, 0.2, 0.7],
leftAdjustAlph=arg[6], ##
rightAdjustAlph=arg[7], ##
Adjustbox=arg[5], ##
pixel_filter=arg[8], ## 参数一是过滤宽，第二个参数是过滤高
batch_by_1=arg[9],
scoremap_enhance_pixel=arg[10],
)
# ti2 = time.asctime(time.localtime(time.time()))
# print("本地时间为 :", ti2)
else:
model = TextOcrModel(ocr, text_detect_p, angle_detect)
rotate_img = color_filter(img, color_thre=255, mode='less')
img = color_filter(rotate_img, color_thre=50, mode='more')
# ------------------------变换---------------------------
gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY)
edges = cv.Canny(gray, 50, 150, apertureSize=3)
# 霍夫变换
lines = cv.HoughLines(edges, 1, np.pi / 180, 0)
for rho, theta in lines[0]:
a = np.cos(theta)
b = np.sin(theta)
x0 = a * rho
y0 = b * rho
x1 = int(x0 + 1000 * (-b))
y1 = int(y0 + 1000 * (a))
x2 = int(x0 - 1000 * (-b))
y2 = int(y0 - 1000 * (a))
# print(x1, '---------', x2, '---------', x2, '---------', y2)
if x1 == x2 or y1 == y2 or y2 == -1000:
rotate_img = img
else:
t = float(y2 - y1) / (x2 - x1)
rotate_angle = math.degrees(math.atan(t))
if rotate_angle > 45:
rotate_angle = -90 + rotate_angle
elif rotate_angle < -45:
rotate_angle = 90 + rotate_angle
rotate_img = ndimage.rotate(img, rotate_angle)
result, angle, im_show = model.model_PSENET(
rotate_img,
scale=arg2[0],
maxScale=arg2[1],
detectAngle=arg2[2], ##是否进行文字方向检测
MAX_HORIZONTAL_GAP=arg2[3], ##字符之间的最大间隔
MIN_V_OVERLAPS=arg2[4],
TEXT_PROPOSALS_MIN_SCORE=arg2[5],
leftAdjustAlph=arg2[6], ##
rightAdjustAlph=arg2[7], ##
Adjustbox=arg2[8], ##
)
# result = union_rbox(result, 0.2)
res = [{
'text': x['text'],
'name': str(i),
'box': {
'cx': x['cx'],
'cy': x['cy'],
'w': x['w'],
'h': x['h'],
'angle': x['degree']
}
} for i, x in enumerate(result)]
res = adjust_box_to_origin(img, angle, res) ##修正box
return res, im_show