class CMND(object):
def __init__(self):
self.Detect_cmnd = CENTER_MODEL(
weight_path="Detect_cmnd/weights/model_cmnd_best.pth")
self.Detect_fields = Detector_fields()
self.ocr = OCR()
self.fields = ['id', 'name', 'date', 'ad1', 'ad2']
def predict(self, img):
restext = {}
t1 = time.time()
font = cv2.FONT_HERSHEY_SIMPLEX
img_aligned = self.Detect_cmnd.detect(img)
print("centernet ", time.time() - t1)
t1 = time.time()
if (img_aligned is not None):
img_aligned = cv2.resize(img_aligned, (800, 650))
res, resimg = self.Detect_fields.detect(img_aligned, 0.3)
print("yolo ", time.time() - t1)
t1 = time.time()
for x in self.fields:
restext[x] = self.ocr.predict(resimg[x], res[x], x)
if (restext[x] is None): restext[x] = ""
print("ocr ", time.time() - t1)
# if(cl=="id"):
# text=(self.id.recognize(im))
# # if(cl=="name"):
# else:
# text=self.str.recognize(im)
# img_aligned=cv2.putText(img_aligned,text,(box[0],box[1]),font,1,(255,0,0),1,cv2.LINE_AA)
#for box in res[x]:
#print("a")
#img_aligned=cv2.rectangle(img_aligned,(box[0],box[1]),(box[2],box[3]),(255,0,0),2,1)
else:
img_aligned = cv2.resize(img, (750, 600))
res, resimg = self.Detect_fields.detect(img_aligned, 0.3)
for x in self.fields:
restext[x] = self.ocr.predict(resimg[x], res[x], x)
#for box in res[x]:
#img_aligned=cv2.rectangle(img_aligned,(box[0],box[1]),(box[2],box[3]),(255,0,0),2,1)
restext['name'] = restext['name'].upper()
return img_aligned, restext
cap = cv.VideoCapture(args.input)
while (cap.isOpened()):
hasFrame, frame = cap.read()
if hasFrame:
roi_imgs = yolo.detect(frame)
index = 0
for roi_img in roi_imgs:
box = [
yolo.bounding_boxes[index][0],
yolo.bounding_boxes[index][1],
yolo.bounding_boxes[index][2],
yolo.bounding_boxes[index][3]
]
pred = ocr.predict(roi_img)
draw_bounding_box(input_image=frame,
bounding_box=box,
label=pred,
background_color=(0, 255, 0),
ocr=ocr)
index += 1
if cv.waitKey(1) & 0xFF == ord('q'):
break
cv.imshow('frame', frame)
out.write(frame)
yolo.clear()
else:
break
parser = argparse.ArgumentParser(description='Testing OCR.')
parser.add_argument('--image', help='Path to image file.')
args = parser.parse_args()
logging.getLogger().setLevel(logging.DEBUG)
# Open the image file
if not os.path.isfile(args.image):
logging.debug("Input image file ", args.image, " doesn't exist")
sys.exit(1)
cap = cv.VideoCapture(args.image)
hasFrame, frame = cap.read()
if hasFrame:
img, characteres = extract_chars(frame,
prefix_label='test_ocr',
min_countours_area_ration=0.01,
debug=True)
#characteres = [img / 255 for img in characteres]
#display_images(characteres, 3, 3)
ocr = OCR(model_filename="../config/emnist_model.pt",
use_cuda=False,
debug=True)
pred = ocr.predict(characteres)
logging.info(f'\nPrediction: {pred}')
else:
logging.debug("Frame not found!")
n += 1
images[str(n)] = img
plot_images(images, row, col, cmap='gray')
if __name__ == '__main__':
parser = argparse.ArgumentParser(description='Testing OCR.')
parser.add_argument('--image', help='Path to image file.')
args = parser.parse_args()
logging.getLogger().setLevel(logging.DEBUG)
# Open the image file
if not os.path.isfile(args.image):
logging.error("Input image file ", args.image, " doesn't exist")
sys.exit(1)
cap = cv.VideoCapture(args.image)
hasFrame, frame = cap.read()
if hasFrame:
images = {}
images['frame'] = frame
ocr = OCR(model_filename="../config/attention_ocr_model.pth", use_cuda=False)
pred = ocr.predict(frame)
logging.info(f'Prediction: {pred}')
plot_images(images, 1, 3, cmap='gray')
else:
logging.debug("Frame not found!")