help='Background subtraction method (KNN, MOG2, COLOR).',
default='COLOR')
parser.add_argument('--train',
type=str,
help='Path to a video or a sequence of image.',
default='data/videos/2.mp4')
args = parser.parse_args()
if args.algo == 'COLOR':
lower_blue = np.array([105, 50, 50])
upper_blue = np.array([130, 255, 255])
lower_black = np.array([95, 45, 45])
upper_black = np.array([140, 255, 255])
detector = Detector(type="COLOR", color=(lower_blue, upper_blue))
tracker = Tracker(160, 30, 10, 100)
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255), (127, 0, 255),
(127, 0, 127)]
#capture = cv.VideoCapture(cv.samples.findFileOrKeep(args.input))
capture = cv.VideoCapture('http://192.168.1.104:8080/video')
if not capture.isOpened:
print('Unable to open: ' + args.input)
exit(0)
frames = 0
inf = 999999991
corners = [[0, 0], [inf, 0], [inf, inf], [0, inf]]
myTable = Table(cellDatabase)
locations = pathCoordinates(dijPath(4, 10, [2, 0], [1, 9]), myTable)
def classifyBoxes(myTable, types):
tracker = Tracker(160, 30, 10, 100)
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
detectorqr = cv.QRCodeDetector()
#capture = cv.VideoCapture(cv.samples.findFileOrKeep(args.input))
capture = cv.VideoCapture('http://192.168.43.1:8080/video')
if not capture.isOpened:
print('Unable to open: ' + args.input)
exit(0)
r = False
frames = 0
inf = 999999991
corners = [[0, 0], [inf, 0], [inf, inf], [0, inf]]
locations = pathCoordinates(dijPath(4, 10, [2, 0], [1, 9]), myTable)
endCells = list(map(myTable.getCellByLocation, locations))
locations = smooth(locations)
index = 0
pastPos = (0, 0) # xpast, ypast
dir = 1 # direction of rotate
hang = 0
hangFrames = 0
endAngle = 90
kp = 0.36
ki = 0.40
kd = 0.4
umax = 150 # max controller output, (N)
alpha = 0.8 # derivative filter smoothing factor
pid = PIDController(kp=kp,
ki=ki,
kd=kd,
max_windup=200,
u_bounds=[-umax, umax],
alpha=alpha)
pid.setTarget(endAngle)
waitingBox = 0
count = 0
t = 0
y0 = 0
y1 = 0
fps = 25
fRotate = 1
cornersList = []
while True:
keyboard = cv.waitKey(1)
if keyboard == 'q' or keyboard == 27:
for i in range(20):
comCells = myTable.getCommonCells(myTable.cells[i])
myTable.cells[i].stop(comCells)
time.sleep(.2)
break
ret, frame = capture.read()
if frame is None:
break
#frame = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)
frames = frames + 1
if frames < 50:
corners1 = getCorners(frame)
cornersList.append(corners1)
continue
if frames < 120:
continue
if frames == 120:
cornersMean = np.mean(cornersList, axis=0)
cornersStd = np.std(cornersList, axis=0)
for i in range(len(cornersList)):
for j in range(len(cornersList[0])):
for k in range(len(cornersList[0][0])):
std = cornersStd[j][k]
mean = cornersMean[j][k]
if (cornersList[i][j][k] < mean - 2 * std) or (
cornersList[i][j][k] > mean + 2 * std):
cornersList[i][j][k] = mean
corners = np.mean(cornersList, axis=0)
frame = getTableFromFrame(corners, frame)
(centers, angles) = detector.Detect(frame)
h1, w1 = frame.shape[:2]
if len(centers) == 0:
continue
print(h1)
print(w1)
if r == False:
data = None
startX = 0
startY = myTable.getCellByLocation([0, 4]).coordinates[0] - 150
[startX, startY] = mmToPixel([startX, startY], w1, h1)
cropFrame = frame[startX:startX + 300, startY:startY + 300]
cv.imshow("cropFrame", cropFrame)
data, bbox, _ = detectorqr.detectAndDecode(cropFrame)
if not data:
continue
if data and r == False:
rightPath = pathCoordinates(dijPath(4, 10, [0, 4], [1, 9]),
myTable)
leftPath = pathCoordinates(dijPath(4, 10, [0, 4], [2, 0]), myTable)
if data == "bshr":
if types[0] == 'left':
locations = leftPath
else:
locations = rightPath
else:
if types[1] == 'right':
locations = rightPath
else:
locations = leftPath
endCells = list(map(myTable.getCellByLocation, locations))
locations = smooth(locations)
print("QR Code detected-->", data)
r = 1
waitingBox = 0
centersMM = pixelToMm((float(centers[0][0]), float(centers[0][1])), w1,
h1)
angle = angles[0][0]
h = [hang, hangFrames, dir]
if index == -1:
cell = myTable.getCellsByNearLocation(centersMM, 1)[0]
if waitingBox:
r = 0
for i in range(20):
comCells = myTable.getCommonCells(myTable.cells[i])
myTable.cells[i].stop(comCells)
time.sleep(.2)
print('waiting a box')
pid = PIDController(kp=kp,
ki=ki,
kd=kd,
max_windup=200,
u_bounds=[-umax, umax],
alpha=alpha)
pid.setTarget(endAngle)
index = 0
t = 0
y1 = 0
fRotate = 1
continue
print('rotating')
if fRotate:
cells = myTable.getCellsByNearLocation(centersMM, 4)
for cell in cells:
myTable.stopCell(cell)
fRotate = 0
w = endAngle - angle
# print(angle, endAngle)
if (abs(w) < 9):
count += 1
if (count > 10):
waitingBox = 1
continue
else:
count = 0
# time step, (sec)
dt = t - pid.last_timestamp_
# Control effort
u = pid.update(y0, t)
y1dot = u
y1 += y1dot * dt
y0 = angle
if u < 0:
u = max(-umax, min(-50, u))
else:
u = max(50, min(umax, u))
# print (u)
myTable.move(cell, 0, 0, -u)
t += 1 / fps
[x, y] = cell.location
continue
[index, hang, hangFrames] = myTable.followPath(locations, centersMM,
angle, index, h)
if hang:
continue
curPos = (centers[0][0], centers[0][1])
[hang, hangFrames, dir] = myTable.isHanging(hang, hangFrames, curPos,
pastPos, dir)
pastPos = curPos
def followshortestPath(myTable, cell1, cell2):
cell1 = [int(i) for i in cell1]
cell2 = [int(i) for i in cell2]
tracker = Tracker(160, 30, 10, 100)
track_colors = [(255, 0, 0), (0, 255, 0), (0, 0, 255), (255, 255, 0),
(0, 255, 255), (255, 0, 255), (255, 127, 255),
(127, 0, 255), (127, 0, 127)]
detectorqr = cv.QRCodeDetector()
#capture = cv.VideoCapture(cv.samples.findFileOrKeep(args.input))
capture = cv.VideoCapture('http://192.168.43.1:8080/video')
if not capture.isOpened:
print('Unable to open: ' + args.input)
exit(0)
r = False
frames = 0
inf = 999999991
corners = [[0, 0], [inf, 0], [inf, inf], [0, inf]]
locations = pathCoordinates(dijPath(4, 10, cell1, cell2), myTable)
endCells = list(map(myTable.getCellByLocation, locations))
locations = smooth(locations)
index = 0
pastPos = (0, 0) # xpast, ypast
dir = 1 # direction of rotate
hang = 0
hangFrames = 0
endAngle = 90
kp = 0.36
ki = 0.40
kd = 0.4
umax = 150 # max controller output, (N)
alpha = 0.8 # derivative filter smoothing factor
pid = PIDController(kp=kp,
ki=ki,
kd=kd,
max_windup=200,
u_bounds=[-umax, umax],
alpha=alpha)
pid.setTarget(endAngle)
waitBox = 0
count = 0
t = 0
y0 = 0
y1 = 0
fps = 25
fRotate = 1
cornersList = []
while True:
keyboard = cv.waitKey(1)
if keyboard == 'q' or keyboard == 27:
for i in range(20):
comCells = myTable.getCommonCells(myTable.cells[i])
myTable.cells[i].stop(comCells)
time.sleep(.2)
break
ret, frame = capture.read()
if frame is None:
break
# frame = cv.resize(frame, (640, 360))
#frame = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)
frames = frames + 1
if frames < 50:
corners1 = getCorners(frame)
cornersList.append(corners1)
continue
if frames < 120:
continue
if frames == 120:
cornersMean = np.mean(cornersList, axis=0)
cornersStd = np.std(cornersList, axis=0)
for i in range(len(cornersList)):
for j in range(len(cornersList[0])):
for k in range(len(cornersList[0][0])):
std = cornersStd[j][k]
mean = cornersMean[j][k]
if (cornersList[i][j][k] < mean - 2 * std) or (
cornersList[i][j][k] > mean + 2 * std):
cornersList[i][j][k] = mean
corners = np.mean(cornersList, axis=0)
frame = getTableFromFrame(corners, frame)
(centers, angles) = detector.Detect(frame)
h1, w1 = frame.shape[:2]
if len(centers) == 0:
continue
if waitBox:
waitBox -= 1
for i in range(20):
comCells = myTable.getCommonCells(myTable.cells[i])
myTable.cells[i].stop(comCells)
continue
centersMM = pixelToMm((float(centers[0][0]), float(centers[0][1])), w1,
h1)
angle = angles[0][0]
h = [hang, hangFrames, dir]
if index == -1:
cell = myTable.getCellsByNearLocation(centersMM, 1)[0]
waitBox = 30
index = 0
[index, hang, hangFrames] = myTable.followPath(locations, centersMM,
angle, index, h)
if hang:
continue
curPos = (centers[0][0], centers[0][1])
[hang, hangFrames, dir] = myTable.isHanging(hang, hangFrames, curPos,
pastPos, dir)
pastPos = curPos