res = []
for j in range(num):
for i in range(meta.classes):
if dets[j].prob[i] > 0:
b = dets[j].bbox
res.append(
(meta.names[i], dets[j].prob[i], (b.x, b.y, b.w, b.h)))
res = sorted(res, key=lambda x: -x[1])
#dn.free_image(im)
#dn.free_detections(dets, num)
return res
if __name__ == "__main__":
dn.srand(2222222)
dn.nnp_initialize()
# Darknet
print("load network and meta(classes and names)")
net = dn.load_net(
b"../cfg/yolov3-tiny.cfg",
b"/home/pi/Libraries/yolo/3/trained_weight/yolov3-tiny.weights", 0)
meta = dn.load_meta(b"../cfg/coco4examples.data")
print("using darknet function")
r = dn.detect(net, meta, b"../data/dog.jpg")
print(r)
# scipy
print("using scipy to pass image to darknet")
arr = imread(b"../data/dog.jpg")
im = array_to_image(arr)
def yolo_detect(self, interval):
time.sleep(interval)
try:
for i in range(24):
self.cap.grab()
ret, image = self.cap.read()
if (cv2.countNonZero(cv2.cvtColor(image,
cv2.COLOR_BGR2GRAY)) == 0):
return False, "Resim Siyah"
else:
image_copy = image.copy()
##################################### Masking ############################################
marginx = 215
marginy = 0
cv2.rectangle(image_copy, (0 + marginx, 0 + marginy),
(50 + marginx, 76 + marginy), (255, 0, 255), -1)
cv2.rectangle(image_copy, (50 + marginx, 0 + marginy),
(100 + marginx, 65 + marginy), (255, 0, 255),
-1) #CAM1
cv2.rectangle(image_copy, (100 + marginx, 0 + marginy),
(150 + marginx, 50 + marginy), (255, 0, 255), -1)
cv2.rectangle(image_copy, (150 + marginx, 0 + marginy),
(210 + marginx, 40 + marginy), (255, 0, 255), -1)
cv2.rectangle(image_copy, (140, 420), (212, 480),
(255, 0, 255), -1) #CAM7
##################################### Masking ############################################
image_dn = self.array_to_image(image_copy)
dn.rgbgr_image(image_dn)
hour = datetime.datetime.now().hour
dn.nnp_initialize()
r = None
#print("inceleniyor")
if (hour > 7 and hour < 18):
r = dn.detect(self.net, self.meta, image_dn, 0.35)
else:
r = dn.detect(self.net, self.meta, image_dn, 0.30)
length = len(r)
#print(r)
flag = False
if (length > 0):
self.DetectedImage = image
for i in range(length):
if (str(r[i][0]).find("person") >= 0):
self.DetectedImage = cv2.rectangle(
self.DetectedImage,
(int(r[i][2][0]) - int(r[i][2][2]) - 5,
int(r[i][2][1]) - int(r[i][2][3]) - 5),
(int(r[i][2][0]) + int(r[i][2][2]) + 5,
int(r[i][2][1]) + int(r[i][2][3]) + 5),
(255, 0, 255), 2)
#print("Detected")
flag = True
return flag, r
except:
#print("error")
return False, "Usb Takili Degil"
def yolo_detect(self, interval):
time.sleep(interval)
try:
for i in range(24):
self.cap.grab()
ret, image = self.cap.read()
cam1 = image[0:360, 0:640]
cam2 = image[0:360, 640:1280]
cam3 = image[0:360, 1280:1920]
cam4 = image[360:720, 0:640]
cam5 = image[360:720, 640:1280]
cam6 = image[360:720, 1280:1920]
cam7 = image[720:1080, 0:640]
cam8 = image[720:1080, 640:1280]
combined_images = np.zeros((720, 1280, 3), np.uint8)
combined_images[0:360, 0:640] = cam1
combined_images[0:360, 640:1280] = cam2
combined_images[360:720, 0:640] = cam4
combined_images[360:720, 640:1280] = cam6
image = combined_images
if (cv2.countNonZero(cv2.cvtColor(image,
cv2.COLOR_BGR2GRAY)) == 0):
return False, "Resim Siyah"
else:
image_copy = image.copy()
##################################### Masking ############################################
cv2.rectangle(image_copy, (600, 300), (650, 370),
(255, 0, 255), -1)
#marginx = 215
#marginy = 0
#cv2.rectangle(image_copy, (0+marginx,0+marginy), (50+marginx, 76+marginy), (255,0,255),-1)
#cv2.rectangle(image_copy, (50+marginx,0+marginy), (100+marginx, 65+marginy), (255,0,255),-1) #CAM1
#cv2.rectangle(image_copy, (100+marginx,0+marginy), (150+marginx, 50+marginy), (255,0,255),-1)
#cv2.rectangle(image_copy, (150+marginx,0+marginy), (210+marginx, 40+marginy), (255,0,255),-1)
#cv2.rectangle(image_copy, (315+marginx,0+marginy), (360+marginx, 75+marginy), (255,0,255),-1) #CAM3 NEW
#cv2.rectangle(image_copy, (140,420), (212, 480), (255,0,255),-1) #CAM7
##################################### Masking ############################################
image_dn = self.array_to_image(image_copy)
dn.rgbgr_image(image_dn)
hour = datetime.datetime.now().hour
dn.nnp_initialize()
r = None
#print("inceleniyor")
if (hour > 7 and hour < 18):
r = dn.detect(self.net, self.meta, image_dn, 0.55)
else:
r = dn.detect(self.net, self.meta, image_dn, 0.45)
length = len(r)
#print(r)
flag = False
if (length > 0):
self.DetectedImage = image
for i in range(length):
if (str(r[i][0]).find("person") >= 0):
self.DetectedImage = cv2.rectangle(
self.DetectedImage,
(int(r[i][2][0]) - int(r[i][2][2]) - 5,
int(r[i][2][1]) - int(r[i][2][3]) - 5),
(int(r[i][2][0]) + int(r[i][2][2]) + 5,
int(r[i][2][1]) + int(r[i][2][3]) + 5),
(255, 0, 255), 2)
#print("Detected")
flag = True
return flag, r
except:
#print("error")
return False, "Usb Takili Degil"