def main():
scores = [0]
angle = 0
iter = 1
start = time.time()
while((np.mean(scores[:3]) < 0.7) and (angle < 180) and (np.mean(scores[:2]) < 0.65) and iter <= 10):
model = load_inference_model(MODEL_PATH)
# load image
image = read_image_bgr(IMAGE_PATH + IMAGE_NAME)
#Rotate
if(iter != 1):
angle = angle + 20
image = rotate(image,angle)
# copy to draw on
draw = image.copy()
draw = cv2.cvtColor(draw, cv2.COLOR_BGR2RGB)
#rotate
# preprocess image for network
image = preprocess_image(image)
image, _ = resize_image(image, 416, 448)
# process image
#start = time.time()
boxes, scores, labels = model.predict_on_batch(np.expand_dims(image, axis=0))
print("iteration: ",str(iter))
print("processing time: ", time.time() - start)
boxes = post_process(boxes, draw, image)
labels = labels[0]
#print(labels)
scores = scores[0]
#print(scores)
boxes = boxes[0]
if(iter == 1):
angle = -40
iter = iter + 1
#print(labels)
#print(scores)
visualize_boxes(draw, boxes, labels, scores, class_labels=['0', '1', '2', '3', '4', '5', '6', '7', '8', '9'])
# 5. plot
plt.imshow(draw)
plt.imsave(SAVE_PATH + IMAGE_NAME,draw)
plt.show()
def processImage(image, index):
Width = image.shape[1]
Height = image.shape[0]
print(Width)
print(Height)
#cv2.imshow('original frame',image)
#cv2.waitKey(0)
#cv2.destroyAllWindows()
# read pre-trained model and config file
net = cv2.dnn.readNet(WEIGHTS, CONFIG)
# create input blob
blob = cv2.dnn.blobFromImage(image,
scale, (416, 416), (0, 0, 0),
True,
crop=False)
#blob = cv2.dnn.blobFromImage(image, scale, (320,320), (0,0,0), True, crop=False)
# set input blob for the network
net.setInput(blob)
# run inference through the network
# and gather predictions from output layers
outs = net.forward(get_output_layers(net))
# initialization
class_ids = []
confidences = []
boxes_b = []
# for each detetion from each output layer
# get the confidence, class id, bounding box params
# and ignore weak detections (confidence < 0.5)
for out in outs:
for detection in out:
scores = detection[5:]
class_id = np.argmax(scores)
confidence = scores[class_id]
if confidence > 0.70:
center_x = int(detection[0] * Width)
center_y = int(detection[1] * Height)
w = int(detection[2] * Width)
h = int(detection[3] * Height)
x = center_x - w / 2
y = center_y - h / 2
class_ids.append(class_id)
confidences.append(float(confidence))
boxes_b.append([x, y, w, h])
# # do OCR in half upper part of the boxes
# print("la parte donde esta mirando el testo de las imagenes");
# cropped_img = image[y:y + h, x:x + w] #--- Notice this part where you have to add the stride as well ---
# cropped_img = cv2.cvtColor(cropped_img, cv2.COLOR_BGR2GRAY) #convert to grey scale
# cropped_img= cv2.bilateralFilter(cropped_img, 11, 17, 17) #Blur to reduce noise
# cv2.imshow('img',cropped_img)
# custom_config = r'--psm 11'
# text = pytesseract.image_to_string(cropped_img,lang="digits_added", config= custom_config)
# print("Cropped:",text)
# cv2.waitKey(0)
# cv2.destroyAllWindows()
# apply non-max suppression
indices = cv2.dnn.NMSBoxes(boxes_b, confidences, conf_threshold,
nms_threshold)
# go through the detections remaining
# after nms and draw bounding box
for i in indices:
i = i[0]
box = boxes_b[i]
x = box[0]
y = box[1]
w = box[2]
h = box[3]
#print("boxes to be drawn");
draw_bounding_box(image, class_ids[i], confidences[i], round(x),
round(y), round(x + w), round(y + h))
# cv2.imshow("object_location_boxes", image)
# # #wait until any key is pressed
# cv2.waitKey()
# cv2.destroyAllWindows()
if (class_ids[i] == 5):
print('bus located')
# do OCR in half upper part of the boxes
print("la parte donde esta mirando el testo de las imagenes")
cropped_img = image[
int(y):int(y) + int(h / 3),
int(x) + int(1 * w / 10):int(x) + int(
7 * w / 10
)] #--- Notice this part where you have to add the stride as well ---
#cropped_img = cv2.cvtColor(cropped_img, cv2.COLOR_BGR2GRAY) #convert to grey scale
#cropped_img= cv2.bilateralFilter(cropped_img, 11, 17, 17) #Blur to reduce noise
cv2.imshow('img', cropped_img)
# copy to draw on
draw = cropped_img.copy()
draw = cv2.cvtColor(draw, cv2.COLOR_BGR2RGB)
# # preprocess image for network
cropped_img = preprocess_image(cropped_img)
cropped_img, _ = resize_image(cropped_img, 416, 448)
# #cropped_img, _ = resize_image(cropped_img, 410, 448)
plt.imshow(cropped_img)
#plt.imshow(_);
# process image
start = time.time()
boxes_d, scores, labels = model.predict_on_batch(
np.expand_dims(cropped_img, axis=0))
print(labels)
print("processing time: ", time.time() - start)
boxes_d = post_process(boxes_d, draw, cropped_img)
labels = labels[0]
scores = scores[0]
boxes_d = boxes_d[0]
print(labels.shape)
print(boxes_d.shape)
print(scores[0])
visualize_boxes(draw,
boxes_d,
labels,
scores,
class_labels=[
'0', '1', '2', '3', '4', '5', '6', '7', '8',
'9'
])
# 5. plot
plt.imshow(draw)
plt.show()
digits_array = np.concatenate(
(boxes_d, scores[:, np.newaxis], labels[:, np.newaxis]),
axis=1)
digits_array = digits_array[digits_array[:, 4] > 0.35]
sortedDigits_score = digits_array[digits_array[:, 0].argsort()]
#sortedDigits_left_right = digits_array[digits_array[:,0].argsort()]
print(sortedDigits_score)
##### UNcomment when I solve the errors when it does not get three numbers and also when you solve the algorithm for knowing how many numbers there are.
# print(sortedDigits_score.shape)
# print("the number is")
# print(str(int(sortedDigits_score[0][5])) + str(int(sortedDigits_score[1][5])) + str(int(sortedDigits_score[2][5])))
# #grayscaled = cv2.cvtColor(cropped_img,cv2.COLOR_BGR2GRAY)
# #brigthness = brigthness_level(grayscaled);
# #print(brigthness);
# #retval, th = cv2.threshold(grayscaled, brigthness*1.2, 255, cv2.THRESH_BINARY_INV)
# #th = cv2.adaptiveThreshold(grayscaled, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 115, 1)
# #retval2,th = cv2.threshold(grayscaled,254,255,cv2.THRESH_BINARY+cv2.THRESH_OTSU)
# #cv2.imshow('Adaptive threshold',th)
# cv2.waitKey(0)
# cv2.destroyAllWindows()
# print("to print image to data");
# print(pytesseract.image_to_data(Image.fromarray(cropped_img), config = '--psm 11',output_type='data.frame'))
# custom_config = r'--psm 11'
# text = pytesseract.image_to_string(cropped_img,lang="digits_added", config= custom_config)
# print("Cropped:",text)
cv2.waitKey(0)
cv2.destroyAllWindows()
# copy to draw on
draw = image.copy()
draw = cv2.cvtColor(draw, cv2.COLOR_BGR2RGB)
# preprocess image for network
image = preprocess_image(image)
image, _ = resize_image(image, 416, 448)
# process image
start = time.time()
boxes, scores, labels = model.predict_on_batch(
np.expand_dims(image, axis=0))
print("processing time: ", time.time() - start)
boxes = post_process(boxes, draw, image)
labels = labels[0]
scores = scores[0]
boxes = boxes[0]
visualize_boxes(
draw,
boxes,
labels,
scores,
class_labels=['0', '1', '2', '3', '4', '5', '6', '7', '8', '9'])
# 5. plot
plt.imshow(draw)
plt.show()