else:
couter = 0
error = 0
if couter > 3:
font = ImageFont.truetype("simsun.ttc", 30, index=1)
img_rd = Image.fromarray(cv2.cvtColor(datum.cvOutputData, cv2.COLOR_BGR2RGB))
draw = ImageDraw.Draw(img_rd)
draw.text((10, 10), text="Fall Detected", font=font,
fill=(255, 0, 0))
img_rd = cv2.cvtColor(np.array(img_rd), cv2.COLOR_RGB2BGR)
time_snap = datetime.datetime.now()
cv2.imwrite('fall_detection' + str(time_snap).replace(':','') + '.jpg', frame)
if (datetime.datetime.now() - pre).total_seconds() > 5:
t = threading.Thread(
target=post(event=3, imagePath='fall_detection' + str(time_snap).replace(':','') + '.jpg'))
t.start()
# status = post(event=3, imagePath='fall_detection.jpg')
# print("fall")
pre = datetime.datetime.now()
# print(pre)
# update variables
frame_start_time = now
v0 = v
width0 = width.copy()
height0 = height.copy()
# if width > height:
# print("alarm")
firstFrame = None
except Exception as e:
def fall_detect(cnts, defined_min_area, frame, prevX, prevY, xList, yList,
centerV, alert, pre):
for c in cnts:
# exclusion
if cv2.contourArea(c) < defined_min_area:
continue
# outer bounding box
(x_b, y_b, w_b, h_b) = cv2.boundingRect(c)
cv2.rectangle(frame, (x_b, y_b), (x_b + w_b, y_b + h_b), (0, 255, 255),
2) # 黄色矩形
# rotating bounding box
rect = cv2.minAreaRect(c) # 得到最小外接矩形的（中心(x,y), (宽,高), 旋转角度）
box = cv2.boxPoints(rect) # 获取最小外接矩形的4个顶点坐标
box = np.int0(box)
cv2.drawContours(frame, [box], 0, (0, 0, 255), 2)
# averaging line
rows, cols = frame.shape[:2]
[vx, vy, x, y] = cv2.fitLine(c, cv2.DIST_L2, 0, 0.01, 0.01)
lefty = (-x * vy / vx) + y
righty = ((cols - x) * vy / vx) + y
cv2.line(frame, (cols - 1, righty), (0, lefty), (255, 0, 0), 2)
# ellipse
elps = cv2.fitEllipse(c)
(x, y), (MA, ma), angle = cv2.fitEllipse(c)
cv2.ellipse(frame, elps, (255, 0, 0), 3) # red
# Aspect Ratio
AR = MA / ma
# Center Speed - acceleration
prevX = 0.0
prevY = 0.0
centerSpeed = 0
if xList.full():
prevX = statistics.median(list(xList.queue))
prevY = statistics.median(list(yList.queue))
xList.get()
yList.get()
xList.put(elps[0][0])
yList.put(elps[0][1])
X = statistics.median(list(xList.queue))
Y = statistics.median(list(yList.queue))
if xList.full():
dx = abs(prevX - X)
dy = abs(prevY - Y)
centerV = math.sqrt(dx**2 + dy**2)
# calculate probabilities for the 4 features
pAngle = (abs(angle - 90) - 50) / 10
pAngle = 1 / (math.exp(pAngle) + 1)
pAR = 10 * AR - 5
pAR = 1 / (math.exp(pAR) + 1)
ACS = centerV - 9
try:
ACS = 1 / (math.exp(ACS) + 1)
except:
ACS = 1 / (float('inf') + 1)
# print("pAngle : ", pAngle)
# print("pAR : ", pAR)
# print("ACS : ", ACS)
# confidence
P_FALL = pAngle * pAR * ACS + 0.5
# print("P_FALL1 : ", P_FALL)
P_FALL = 1 / (math.exp(-(P_FALL - 0.65) * 10) + 1)
# print("P_FALL2: ", P_FALL)
# status display
# cv2.putText(frame, "Status : ", (10, 20), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 14)
# cv2.putText(frame, "Fall Confidence: {0:.2f} ".format(P_FALL), (10, 50), cv2.FONT_HERSHEY_SIMPLEX, 0.5,
# (0, 128, 255), 14)
# cv2.putText(frame, "Angle: {0:.2f}".format(angle), (10, 220),cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 0), 14)
# cv2.putText(frame, "AR: {0:.2f}".format(AR), (10, 237),cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 0), 14)
# cv2.putText(frame, "Center Speed: {0:.2f}".format(centerV), (10, 256),cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 0), 14)
# fall
if P_FALL > 0.88:
if alert > 3:
# print("fall")
font = ImageFont.truetype("simsun.ttc", 30, index=1)
img_rd = Image.fromarray(cv2.cvtColor(frame,
cv2.COLOR_BGR2RGB))
draw = ImageDraw.Draw(img_rd)
draw.text((10, 10),
text="Fall Detected",
font=font,
fill=(255, 0, 0))
frame = cv2.cvtColor(np.array(img_rd), cv2.COLOR_RGB2BGR)
time_snap = datetime.datetime.now()
cv2.imwrite(
'fall_detection' + str(time_snap).replace(':', '') +
'.jpg', frame)
if (datetime.datetime.now() - pre).total_seconds() > 5:
t = threading.Thread(
target=post(event=3,
imagePath='fall_detection' +
str(time_snap).replace(':', '') + '.jpg'))
t.setDaemon(False)
t.start()
pre = datetime.datetime.now()
# cv2.imwrite("report.jpg", frame)
# send_alert.SendMail("report.jpg")
alert = alert + 1
else:
alert = alert + 1
return frame, alert, pre
def tick(checkItem, url):
data = checkItem.check()
post(data, url)
def process(self, img_rd):
img_with_name = img_rd
data_type_three = {
'old': 0,
'employee': 0,
'volunteer': 0,
'stranger': 0
}
# 读取所有人脸
if self.get_face_database():
cv2.putText(img_rd, "Faces: " + str(self.faces_cnt), (20, 40), cv2.FONT_ITALIC, 0.8, (0, 255, 0), 1,
cv2.LINE_AA)
# cv2.putText(img_rd, str(datetime.now()), (120, 40), cv2.FONT_ITALIC, 0.8, (0, 255, 0), 1,
# cv2.LINE_AA)
self.features_camera_list = []
self.faces_cnt = 0
self.pos_camera_list = []
self.name_camera_list = []
self.type_camera_list = []
self.id_camera_list = []
(h, w) = img_rd.shape[:2]
blob = cv2.dnn.blobFromImage(cv2.resize(img_rd, (300, 300)), 1.0,
(300, 300), (104.0, 177.0, 123.0))
self.detector.setInput(blob)
faces = self.detector.forward()
# 检测到人脸
if faces.shape[2] != 0:
# 遍历捕获到的图像中所有的人脸
for k in range(0, faces.shape[2]):
# 计算矩形框大小
confidence = faces[0, 0, k, 2]
# filter out weak detections by ensuring the `confidence` is
# greater than the minimum confidence
if confidence < 0.5:
continue
self.faces_cnt += 1
# 让人名跟随在矩形框的上方
# 确定人名的位置坐标
# 先默认所有人不认识，是 unknown
self.name_camera_list.append("unknown")
self.type_camera_list.append('unknown')
self.id_camera_list.append('unknown')
# 每个捕获人脸的名字坐标
box = faces[0, 0, k, 3:7] * np.array([w, h, w, h])
(startX, startY, endX, endY) = box.astype("int")
self.pos_camera_list.append(tuple(
[int(startX + 5), int(startY - 30)]))
# height = (endY - startY)
# width = (endX - startX)
img_blank = img_rd[startY:endY, startX:endX]
img_blank = img_blank[..., ::-1]
try:
# for ii in range(height):
# for jj in range(width):
# img_blank[ii][jj] = img_rd[startY + ii][startX + jj]
img = cv2.resize(img_blank, (96, 96))
img = (img / 255.).astype(np.float32)
img = self.nn4_small2.predict(np.expand_dims(img, axis=0))[0]
# 对于某张人脸，遍历所有存储的人脸
e_distance_list = []
for i in range(0, len(self.embedded)):
e_distance_list.append(facenet.distance(self.embedded[i], img))
similar_person_num = e_distance_list.index(min(e_distance_list))
# print(min(e_distance_list))
if min(e_distance_list) < 0.58:
self.name_camera_list[k] = self.id_known_list[similar_person_num % 8]
self.type_camera_list[k] = self.type_known_list[similar_person_num % 8]
self.id_camera_list[k] = self.name_known_list[similar_person_num % 8]
data_type_three[api_transfer[self.type_camera_list[k]]] += 1
cv2.rectangle(img_rd, tuple([startX, startY]), tuple([endX, endY]),
(0, 255, 0), 2)
cv2.rectangle(img_rd, tuple([startX, startY - 35]), tuple([endX, startY]),
(0, 255, 0), cv2.FILLED)
img_with_name = self.draw_name(img_rd)
if self.type_camera_list[k] == 'elder':
mode = smile_detection.smile_detect(img_blank)
if mode == 'happy':
# print("happy")
cv2.rectangle(img_with_name, tuple([startX, startY - 70]),
tuple([endX, startY - 35]),
(0, 215, 255), cv2.FILLED)
cv2.putText(img_with_name, 'happy', (startX + 5, startY - 45), cv2.FONT_ITALIC, 1,
(255, 255, 255), 1, cv2.LINE_AA)
time_snap = datetime.now()
cv2.imwrite('smile_detection' + str(time_snap).replace(':','') + '.jpg', img_with_name)
if (datetime.now() - self.pre).total_seconds() > 5:
t = threading.Thread(target=post(elder_id=self.id_camera_list[k], event=0,
imagePath='smile_detection' + str(
time_snap).replace(':','') + '.jpg'))
t.setDaemon(False)
t.start()
self.pre = datetime.now()
# print("May be person " + str(self.name_known_list[similar_person_num]))
elif min(e_distance_list) > 0.75:
data_type_three['stranger'] += 1
self.name_camera_list[k] = '陌生人'
cv2.rectangle(img_rd, tuple([startX, startY]), tuple([endX, endY]),
(0, 0, 255), 2)
cv2.rectangle(img_rd, tuple([startX, startY - 35]), tuple([endX, startY]),
(0, 0, 255), cv2.FILLED)
img_with_name = self.draw_name(img_rd)
time_snap = datetime.now()
cv2.imwrite('stranger_detection' + str(time_snap).replace(':','') + '.jpg', img_with_name)
if (datetime.now() - self.pre).total_seconds() > 5:
t = threading.Thread(
target=post(event=2, imagePath='stranger_detection' + str(time_snap).replace(':','') + '.jpg'))
t.setDaemon(False)
t.start()
self.pre = datetime.now()
else:
pass
except:
continue
else:
img_with_name = img_rd
# 更新 FPS / Update stream FPS
# self.update_fps()
if (datetime.now() - self.pre).total_seconds() > 5:
post_person(data_type_three)
self.pre = datetime.now()
return img_with_name
def tick(checkItem):
data = checkItem.check()
post(data, '/checkData')
def run(self, frame):
height = []
width = []
center = []
# Process Image
datum = op.Datum()
datum.cvInputData = frame
self.opWrapper.emplaceAndPop([datum])
img_rd = datum.cvOutputData
# fall judge
try:
# key points have been identified
x = datum.poseKeypoints[0][:, 0]
y = datum.poseKeypoints[0][:, 1]
width.append(np.max(x[np.nonzero(x)]) - np.min(x[np.nonzero(x)]))
height.append(np.max(y[np.nonzero(y)]) - np.min(y[np.nonzero(y)]))
center.append(np.mean(x[np.nonzero(x)]))
center.append(np.mean(y[np.nonzero(y)]))
if self.frame_start_time == 0:
self.center0 = center.copy()
self.width0 = width.copy()
self.height0 = height.copy()
self.frame_start_time = time.time()
else:
diff = np.array(
[center[0] - self.center0[0], center[1] - self.center0[1]])
dist = math.sqrt(np.sum((diff * 10**(-4))**2))
now = time.time()
v = dist / (now - self.frame_start_time)
a = (v**2 - self.v0**2) / (2 * dist)
# print(v, abs(a))
if (abs(a) > 0.2) and \
(np.subtract(np.array(width), np.array(height)) > np.subtract(np.array(self.width0), np.array(
self.height0)) and np.subtract(np.array(width), np.array(height)) > 0):
self.couter += 1
# print("alarm by v and a")
elif (width > height and (x[8] != 0 or x[9] != 0 or x[12] != 0)
and v < 1):
self.couter += 1
# print("alarm by w and h")
else:
if self.error == 0:
self.error += 1
else:
self.couter = 0
self.error = 0
if self.couter > 3:
font = ImageFont.truetype("simsun.ttc", 30, index=1)
img_rd = Image.fromarray(
cv2.cvtColor(datum.cvOutputData, cv2.COLOR_BGR2RGB))
draw = ImageDraw.Draw(img_rd)
draw.text((10, 10),
text="Fall Detected",
font=font,
fill=(255, 0, 0))
img_rd = cv2.cvtColor(np.array(img_rd), cv2.COLOR_RGB2BGR)
cv2.imwrite('fall_detection.jpg', frame)
t = threading.Thread(
target=post(event=3, imagePath='fall_detection.jpg'))
t.setDaemon(False)
t.start()
# status = post(event=3, imagePath='fall_detection.jpg')
# print("fall")
# update variables
self.frame_start_time = now
self.v0 = v
self.width0 = width.copy()
self.height0 = height.copy()
# if width > height:
# print("alarm")
self.firstFrame = None
except Exception as e:
gray = frame_process.preprocess_frame(frame)
if self.firstFrame is None:
self.firstFrame = gray
pass
frameDelta = cv2.absdiff(self.firstFrame, gray)
cnts = frame_process.get_contours(self.firstFrame, gray)
defined_min_area = 3000
frame, self.alert = algorithm_fall.fall_detect(
cnts, defined_min_area, frame, self.prevX, self.prevY,
self.xList, self.yList, self.centerV, self.alert)
img_rd = frame
# cv2.imshow("OpenPose 1.6.0 - Tutorial Python API", frame)
frame = cv2.resize(img_rd, (640, 480))
# cv2.imshow("OpenPose 1.6.0 - Tutorial Python API", img_rd)
return frame
# http://zhuooyu.cn:8000/api/person/old/10
def process(self, frame):
# grab the next frame and handle if we are reading from either
# VideoCapture or VideoStream
rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
# if the frame dimensions are empty, set them
if self.W is None or self.H is None:
(self.H, self.W) = frame.shape[:2]
# initialize the current status along with our list of bounding
# box rectangles returned by either (1) our object detector or
# (14) the correlation trackers
status = "Waiting"
rects = []
# check to see if we should run a more computationally expensive
# object detection method to aid our tracker
if self.totalFrames % 20 == 0:
# set the status and initialize our new set of object trackers
status = "Detecting"
self.trackers = []
# convert the frame to a blob and pass the blob through the
# network and obtain the detections
blob = cv2.dnn.blobFromImage(frame, 0.007843, (self.W, self.H), 127.5)
self.net.setInput(blob)
detections = self.net.forward()
# loop over the detections
for i in np.arange(0, detections.shape[2]):
# extract the confidence (i.e., probability) associated
# with the prediction
confidence = detections[0, 0, i, 2]
# filter out weak detections by requiring a minimum
# confidence
if confidence > 0.5:
# extract the index of the class label from the
# detections list
idx = int(detections[0, 0, i, 1])
# if the class label is not a person, ignore it
if CLASSES[idx] != "person":
continue
# compute the (x, y)-coordinates of the bounding box
# for the object
box = detections[0, 0, i, 3:7] * np.array([self.W, self.H, self.W, self.H])
(startX, startY, endX, endY) = box.astype("int")
# construct a dlib rectangle object from the bounding
# box coordinates and then start the dlib correlation
# tracker
tracker = dlib.correlation_tracker()
rect = dlib.rectangle(startX, startY, endX, endY)
tracker.start_track(rgb, rect)
# add the tracker to our list of trackers so we can
# utilize it during skip frames
self.trackers.append(tracker)
# otherwise, we should utilize our object *trackers* rather than
# object *detectors* to obtain a higher frame processing throughput
else:
# loop over the trackers
for tracker in self.trackers:
# set the status of our system to be 'tracking' rather
# than 'waiting' or 'detecting'
status = "Tracking"
# update the tracker and grab the updated position
tracker.update(rgb)
pos = tracker.get_position()
# unpack the position object
startX = int(pos.left())
startY = int(pos.top())
endX = int(pos.right())
endY = int(pos.bottom())
# draw a rectangle around the people
cv2.rectangle(frame, (startX, startY), (endX, endY),
(0, 255, 0), 2)
# add the bounding box coordinates to the rectangles list
rects.append((startX, startY, endX, endY))
# draw a horizontal line in the center of the frame -- once an
# object crosses this line we will determine whether they were
# moving 'up' or 'down'
# cv2.line(frame, (0, H // 14), (W, H // 14), (0, 255, 255), 14)
# use the centroid tracker to associate the (1) old object
# centroids with (14) the newly computed object centroids
objects = self.ct.update(rects)
# loop over the tracked objects
for (objectID, centroid) in objects.items():
# check to see if a trackable object exists for the current
# object ID
to = self.trackableObjects.get(objectID, None)
# if there is no existing trackable object, create one
if to is None:
to = TrackableObject(objectID, centroid)
# otherwise, there is a trackable object so we can utilize it
# to determine direction
else:
# the difference between the y-coordinate of the *current*
# centroid and the mean of *previous* centroids will tell
# us in which direction the object is moving (negative for
# 'up' and positive for 'down')
y = [c[1] for c in to.centroids]
direction = centroid[1] - np.mean(y)
to.centroids.append(centroid)
# check to see if the object has been counted or not
if not to.counted:
# if the direction is negative (indicating the object
# is moving up) AND the centroid is above the center
# line, count the object
if direction < 0 and centroid[1] < self.H // 2:
self.totalUp += 1
to.counted = True
# if the direction is positive (indicating the object
# is moving down) AND the centroid is below the
# center line, count the object
elif direction > 0 and centroid[1] > self.H // 2:
self.totalDown += 1
to.counted = True
current_time = time.strftime('%Y-%m-%d %H:%M:%S',
time.localtime(time.time()))
event_desc = '有人闯入禁止区域!!!'
event_location = '院子'
print('[EVENT] %s, 院子, 有人闯入禁止区域!!!'
% (current_time))
time_snap = datetime.now()
cv2.imwrite('intrusion' + str(time_snap).replace(':', '') + '.jpg', frame)
if (datetime.now() - self.pre).total_seconds() > 5:
t = threading.Thread(
target=post(event=4, imagePath='intrusion' + str(time_snap).replace(':', '') + '.jpg'))
t.setDaemon(False)
t.start()
self.pre = datetime.now()
# todo insert into database
# command = '%s inserting.py --event_desc %s--event_type4 - -event_location % s' % \
# (python_path, event_desc, event_location)
# p = subprocess.Popen(command, shell=True)
# store the trackable obj ect in our dictionary
self.trackableObjects[objectID] = to
# draw both the ID of the object and the centroid of the
# object on the output frame
text = "ID {}".format(objectID)
cv2.putText(frame, text, (centroid[0] - 10, centroid[1] - 10),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 255, 0), 2)
cv2.circle(frame, (centroid[0], centroid[1]), 4,
(0, 255, 0), -1)
# construct a tuple of information we will be displaying on the
# frame
info = [
# ("Up", totalUp),
# ("Down", totalDown),
("Status", status),
]
# loop over the info tuples and draw them on our frame
for (i, (k, v)) in enumerate(info):
text = "{}: {}".format(k, v)
cv2.putText(frame, text, (10, self.H - ((i * 20) + 20)),
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 2)
# show the output frame
frame = cv2.resize(frame, (640, 480))
# increment the total number of frames processed thus far and
# then update the FPS counter
self.totalFrames += 1
return frame
def process(self, img_rd, scale):
img_with_name = img_rd
data_type_three = {
'old': 0,
'employee': 0,
'volunteer': 0,
'stranger': 0
}
# 读取所有人脸
if self.get_face_database():
self.draw_note(img_rd)
self.features_camera_list = []
self.faces_cnt = 0
self.pos_camera_list = []
self.name_camera_list = []
self.type_camera_list = []
(h, w) = img_rd.shape[:2]
blob = cv2.dnn.blobFromImage(cv2.resize(img_rd, (300, 300)), 1.0,
(300, 300), (104.0, 177.0, 123.0))
self.detector.setInput(blob)
faces = self.detector.forward()
# 检测到人脸
if faces.shape[2] != 0:
# 遍历捕获到的图像中所有的人脸
for k in range(0, faces.shape[2]):
# 计算矩形框大小
confidence = faces[0, 0, k, 2]
# filter out weak detections by ensuring the `confidence` is
# greater than the minimum confidence
if confidence < 0.5:
continue
self.faces_cnt += 1
# 让人名跟随在矩形框的上方
# 确定人名的位置坐标
# 先默认所有人不认识，是 unknown
self.name_camera_list.append("unknown")
self.type_camera_list.append("unknown")
# 每个捕获人脸的名字坐标
box = faces[0, 0, k, 3:7] * np.array([w, h, w, h])
(startX, startY, endX, endY) = box.astype("int")
self.pos_camera_list.append(tuple(
[int(startX + 5), int(startY - 30)]))
img_blank = img_rd[startY:endY, startX:endX]
img_blank = img_blank[..., ::-1]
try:
# for ii in range(height):
# for jj in range(width):
# img_blank[ii][jj] = img_rd[startY + ii][startX + jj]
img = cv2.resize(img_blank, (96, 96))
img = (img / 255.).astype(np.float32)
img = self.nn4_small2.predict(np.expand_dims(img, axis=0))[0]
# 对于某张人脸，遍历所有存储的人脸
e_distance_list = []
for i in range(0, len(self.embedded)):
e_distance_list.append(facenet.distance(self.embedded[i], img))
similar_person_num = e_distance_list.index(min(e_distance_list))
# print(min(e_distance_list))
if min(e_distance_list) < 0.58:
self.name_camera_list[k] = self.name_known_list[similar_person_num % 8]
self.type_camera_list[k] = self.type_known_list[similar_person_num % 8]
data_type_three[api_transfer[self.type_camera_list[k]]] += 1
# 绘制矩形框
if self.name_camera_list[k] != 'unknown':
cv2.rectangle(img_rd, tuple([startX, startY]), tuple([endX, endY]),
(0, 255, 0), 2)
cv2.rectangle(img_rd, tuple([startX, startY - 35]), tuple([endX, startY]),
(0, 255, 0), cv2.FILLED)
except:
continue
print(self.type_camera_list)
img_with_name = self.draw_name(img_rd)
if 'unknown' in self.type_camera_list and len(self.type_camera_list) > 1:
index = self.type_camera_list.index('unknown')
pos_vol = np.array(self.pos_camera_list[index])
for i in range(0, len(self.type_camera_list)):
if self.type_camera_list[i] == "elder":
d = scale * np.sqrt(facenet.distance(pos_vol, np.array(self.pos_camera_list[i])))
if d < 50:
if (datetime.now() - self.pre).total_seconds() > 5:
time_snap = datetime.now()
cv2.imwrite("interaction"+str(time_snap).replace(':','')+".jpg", img_with_name)
t = threading.Thread(target=post(event=1, imagePath='interaction'+str(time_snap).replace(':','')+'.jpg'))
t.setDaemon(False)
t.start()
self.pre = datetime.now()
else:
img_with_name = img_rd
# if 'volunteer' in self.type_camera_list:
# 更新 FPS / Update stream FPS
self.update_fps()
# 距离检测
if (datetime.now() - self.pre).total_seconds() > 5:
post_person(data_type_three)
self.pre = datetime.now()
return img_with_name
