def binaryMask(frame, x0, y0, width, height, framecount, plot):
global guessGesture, visualize, mod, saveImg
cv2.rectangle(frame, (x0, y0), (x0 + width, y0 + height), (0, 255, 0), 1)
roi = frame[y0:y0 + height, x0:x0 + width]
gray = cv2.cvtColor(roi, cv2.COLOR_BGR2GRAY)
blur = cv2.GaussianBlur(gray, (5, 5), 2)
th3 = cv2.adaptiveThreshold(blur, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
cv2.THRESH_BINARY_INV, 11, 2)
ret, res = cv2.threshold(th3, minValue, 255,
cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)
if saveImg == True:
saveROIImg(res)
elif guessGesture == True and (framecount % 5) == 4:
t = threading.Thread(target=myNN.guessGesture, args=[mod, res])
t.start()
elif visualize == True:
layer = int(input("Enter which layer to visualize "))
cv2.waitKey(1)
myNN.visualizeLayers(mod, res, layer)
visualize = False
return res
def bkgrndSubMask(frame, x0, y0, width, height, framecount, plot):
global guessGesture, takebkgrndSubMask, visualize, mod, bkgrnd, saveImg
cv2.rectangle(frame, (x0, y0), (x0 + width, y0 + height), (0, 255, 0), 1)
roi = frame[y0:y0 + height, x0:x0 + width]
roi = cv2.cvtColor(roi, cv2.COLOR_BGR2GRAY)
if takebkgrndSubMask == True:
bkgrnd = roi
takebkgrndSubMask = False
print("Refreshing background image for mask...")
diff = cv2.absdiff(roi, bkgrnd)
_, diff = cv2.threshold(diff, 25, 255, cv2.THRESH_BINARY)
mask = cv2.GaussianBlur(diff, (3, 3), 5)
mask = cv2.erode(diff, skinkernel, iterations=1)
mask = cv2.dilate(diff, skinkernel, iterations=1)
res = cv2.bitwise_and(roi, roi, mask=mask)
if saveImg == True:
saveROIImg(res)
elif guessGesture == True and (framecount % 5) == 4:
t = threading.Thread(target=myNN.guessGesture, args=[mod, res])
t.start()
elif visualize == True:
layer = int(input("Enter which layer to visualize "))
cv2.waitKey(0)
myNN.visualizeLayers(mod, res, layer)
visualize = False
return res
def skinMask(frame, x0, y0, width, height, framecount, plot):
global guessGesture, visualize, mod, saveImg
# HSV values
low_range = np.array([0, 50, 80])
upper_range = np.array([30, 200, 255])
cv2.rectangle(frame, (x0, y0), (x0 + width, y0 + height), (0, 255, 0), 1)
roi = frame[y0:y0 + height, x0:x0 + width]
hsv = cv2.cvtColor(roi, cv2.COLOR_BGR2HSV)
mask = cv2.inRange(hsv, low_range, upper_range)
mask = cv2.erode(mask, skinkernel, iterations=1)
mask = cv2.dilate(mask, skinkernel, iterations=1)
mask = cv2.GaussianBlur(mask, (15, 15), 1)
res = cv2.bitwise_and(roi, roi, mask=mask)
# color to grayscale
res = cv2.cvtColor(res, cv2.COLOR_BGR2GRAY)
if saveImg == True:
saveROIImg(res)
elif guessGesture == True and (framecount % 5) == 4:
t = threading.Thread(target=myNN.guessGesture, args=[mod, res])
t.start()
elif visualize == True:
layer = int(input("Enter which layer to visualize "))
cv2.waitKey(0)
myNN.visualizeLayers(mod, res, layer)
visualize = False
return res
def binaryMask(frame, x0, y0, width, height):
global guessGesture, visualize, mod, lastgesture, saveImg
cv2.rectangle(frame, (x0, y0), (x0 + width, y0 + height), (0, 255, 0), 1)
roi = frame[y0:y0 + height, x0:x0 + width]
gray = cv2.cvtColor(roi, cv2.COLOR_BGR2GRAY)
blur = cv2.GaussianBlur(gray, (5, 5), 2)
th3 = cv2.adaptiveThreshold(blur, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
cv2.THRESH_BINARY_INV, 11, 2)
ret, res = cv2.threshold(th3, minValue, 255,
cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)
guess_res = "Bin"
if saveImg == True:
saveROIImg(res)
elif guessGesture == True:
retgesture, guess_res = myNN.guessGesture(mod, res)
if lastgesture != retgesture:
lastgesture = retgesture
elif visualize == True:
layer = int(raw_input("Enter which layer to visualize "))
cv2.waitKey(1)
myNN.visualizeLayers(mod, res, layer)
visualize = False
return res, guess_res
def skinMask(frame, x0, y0, width, height):
global guessGesture, visualize, mod, lastgesture, saveImg
# HSV values
low_range = np.array([0, 50, 80])
upper_range = np.array([30, 200, 255])
cv2.rectangle(frame, (x0, y0), (x0 + width, y0 + height), (0, 255, 0), 1)
roi = frame[y0:y0 + height, x0:x0 + width]
hsv = cv2.cvtColor(roi, cv2.COLOR_BGR2HSV)
#Apply skin color range
mask = cv2.inRange(hsv, low_range, upper_range)
mask = cv2.erode(mask, skinkernel, iterations=1)
mask = cv2.dilate(mask, skinkernel, iterations=1)
#blur
mask = cv2.GaussianBlur(mask, (15, 15), 1)
#cv2.imshow("Blur", mask)
#bitwise and mask original frame
res = cv2.bitwise_and(roi, roi, mask=mask)
# color to grayscale
res = cv2.cvtColor(res, cv2.COLOR_BGR2GRAY)
retgesture = "^______^Skin"
if saveImg == True:
saveROIImg(res)
elif guessGesture == True:
retgesture = myNN.guessGesture(mod, res)
if lastgesture != retgesture:
lastgesture = retgesture
print myNN.output[lastgesture]
time.sleep(0.01)
#guessGesture = False
elif visualize == True:
layer = int(raw_input("Enter which layer to visualize "))
cv2.waitKey(0)
myNN.visualizeLayers(mod, res, layer)
visualize = False
return res, retgesture
def binaryMask(frame, x0, y0, width, height):
global guessGesture, visualize, mod, lastgesture, saveImg
cv2.rectangle(frame, (x0, y0), (x0 + width, y0 + height), (0, 255, 0), 1)
roi = frame[y0:y0 + height, x0:x0 + width]
gray = cv2.cvtColor(roi, cv2.COLOR_BGR2GRAY)
blur = cv2.GaussianBlur(gray, (5, 5), 2)
#blur = cv2.bilateralFilter(roi,9,75,75)
th3 = cv2.adaptiveThreshold(blur, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
cv2.THRESH_BINARY_INV, 11, 2)
ret, res = cv2.threshold(th3, minValue, 255,
cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)
#ret, res = cv2.threshold(blur, minValue, 255, cv2.THRESH_BINARY +cv2.THRESH_OTSU)
retgesture = "^______^Bin"
if saveImg == True:
saveROIImg(res)
elif guessGesture == True:
retgesture = myNN.guessGesture(mod, res)
if lastgesture != retgesture:
lastgesture = retgesture
#print lastgesture
## Checking for only PUNCH gesture here
## Run this app in Prediction Mode and keep Chrome browser on focus with Internet Off
## And have fun :) with Dino
if lastgesture == 3:
jump = ''' osascript -e 'tell application "System Events" to key code 49' '''
#jump = ''' osascript -e 'tell application "System Events" to key down (49)' '''
os.system(jump)
print myNN.output[lastgesture] + "= Dino JUMP!"
# time.sleep(0.01 )
# guessGesture = False
elif visualize == True:
layer = int(raw_input("Enter which layer to visualize "))
cv2.waitKey(1)
myNN.visualizeLayers(mod, res, layer)
visualize = False
return res, retgesture
def bkgrndSubMask(frame, x0, y0, width, height, framecount, plot):
global guessGesture, takebkgrndSubMask, visualize, mod, bkgrnd, lastgesture, saveImg, drawSquare
if drawSquare:
cv2.rectangle(frame, (x0, y0), (x0 + width, y0 + height), (0, 255, 0),
1)
roi = frame[y0:y0 + height, x0:x0 + width]
#roi = cv2.UMat(frame[y0:y0+height, x0:x0+width])
roi = cv2.cvtColor(roi, cv2.COLOR_BGR2GRAY)
#Take background image
if takebkgrndSubMask == True:
bkgrnd = roi
takebkgrndSubMask = False
print("Refreshing background image for mask...")
#Take a diff between roi & bkgrnd image contents
diff = cv2.absdiff(roi, bkgrnd)
#was 25
_, diff = cv2.threshold(diff, 25, 255, cv2.THRESH_BINARY)
# mask = cv2.GaussianBlur(diff, (3,3), 5)
# mask = cv2.erode(diff, skinkernel, iterations = 1)
mask = cv2.dilate(diff, skinkernel, iterations=1)
centerX, topY = 200, 200
try:
centerX = int(np.mean(np.nonzero(diff)[1]))
topY = int(np.percentile(np.nonzero(diff)[0], 5))
except:
pass #lol
res = cv2.bitwise_and(roi, roi, mask=mask)
# res = diff
if saveImg == True:
saveROIImg(res)
elif guessGesture == True and (framecount % 5) == 4:
t = threading.Thread(target=myNN.guessGesture, args=[mod, res])
t.start()
#t.join()
#myNN.update(plot)
elif visualize == True:
layer = int(input("Enter which layer to visualize "))
cv2.waitKey(0)
myNN.visualizeLayers(mod, res, layer)
visualize = False
return res, centerX, topY
def adaptiveSubMask(frame, x0, y0, width, height, framecount, plot):
global guessGesture, takebkgrndSubMask, visualize, mod, bkgrnd, lastgesture, saveImg, lastFrame, avgFrame
cv2.rectangle(frame, (x0, y0), (x0 + width, y0 + height), (0, 255, 0), 1)
roi = frame[y0:y0 + height, x0:x0 + width]
# roi = cv2.UMat(frame[y0:y0+height, x0:x0+width])
roi = cv2.cvtColor(roi, cv2.COLOR_BGR2GRAY)
if (lastFrame is None):
lastFrame = roi
avgFrame = roi
percentNew = .2
avgFrame = avgFrame * (1 - percentNew) + roi * percentNew
bkgrnd = avgFrame.astype('uint8')
# Take background image
# if takebkgrndSubMask == True:
# bkgrnd = roi
# takebkgrndSubMask = False
# print("Refreshing background image for mask...")
# Take a diff between roi & bkgrnd image contents
diff = cv2.absdiff(roi, bkgrnd)
# was 25
_, diff = cv2.threshold(diff, 25, 255, cv2.THRESH_BINARY)
# mask = cv2.GaussianBlur(diff, (3,3), 5)
# mask = cv2.erode(diff, skinkernel, iterations = 1)
mask = cv2.dilate(diff, skinkernel, iterations=1)
res = cv2.bitwise_and(roi, roi, mask=mask)
# res = diff
if saveImg == True:
saveROIImg(res)
elif guessGesture == True and (framecount % 5) == 4:
t = threading.Thread(target=myNN.guessGesture, args=[mod, res])
t.start()
# t.join()
# myNN.update(plot)
elif visualize == True:
layer = int(input("Enter which layer to visualize "))
cv2.waitKey(0)
myNN.visualizeLayers(mod, res, layer)
visualize = False
lastFrame = roi
return res
def skinMask(frame, plot):
global guessGesture, visualize, mod, lastgesture, saveImg
## HSV values
low_range = np.array([0, 50, 80])
upper_range = np.array([30, 200, 255])
cv2.rectangle(frame, (220, 1), (420, 201), (0, 255, 0), 1)
#roi = cv2.UMat(frame[y0:y0+height, x0:x0+width])
roi = frame[1:201, 220:420]
hsv = cv2.cvtColor(roi, cv2.COLOR_BGR2HSV)
## Apply skin color range
mask = cv2.inRange(hsv, low_range, upper_range)
mask = cv2.erode(mask, skinkernel, iterations=1)
mask = cv2.dilate(mask, skinkernel, iterations=1)
## blur
mask = cv2.GaussianBlur(mask, (15, 15), 1)
## bitwise and mask original frame
res = cv2.bitwise_and(roi, roi, mask=mask)
## color to grayscale
res = cv2.cvtColor(res, cv2.COLOR_BGR2GRAY)
if saveImg == True:
saveROIImg(res)
elif guessGesture == True:
#res = cv2.UMat.get(res)
t = threading.Thread(target=myNN.guessGesture, args=[mod, res])
t.start()
elif visualize == True:
layer = int(raw_input("Enter which layer to visualize "))
cv2.waitKey(0)
myNN.visualizeLayers(mod, res, layer)
visualize = False
return res
def Main():
global guessGesture, visualize, mod, binaryMode, bkgrndSubMode, mask, takebkgrndSubMask, x0, y0, width, height, saveImg, gestname, path
quietMode = False
font = cv2.FONT_HERSHEY_SIMPLEX
size = 0.5
fx = 10
fy = 350
fh = 18
#Call CNN model loading callback
while True:
ans = int(input(banner))
if ans == 2:
mod = myNN.loadCNN(-1)
myNN.trainModel(mod)
input("Press any key to continue")
break
elif ans == 1:
print("Will load default weight file")
mod = myNN.loadCNN(0)
break
elif ans == 3:
if not mod:
w = int(input("Which weight file to load (0 or 1)"))
mod = myNN.loadCNN(w)
else:
print("Will load default weight file")
img = int(input("Image number "))
layer = int(input("Enter which layer to visualize "))
myNN.visualizeLayers(mod, img, layer)
input("Press any key to continue")
continue
else:
print("Get out of here!!!")
return 0
## Grab camera input
cap = cv2.VideoCapture(0)
cv2.namedWindow('Original', cv2.WINDOW_NORMAL)
# set rt size as 640x480
ret = cap.set(3, 640)
ret = cap.set(4, 480)
framecount = 0
fps = ""
start = time.time()
plot = np.zeros((512, 512, 3), np.uint8)
global x
while (True):
ret, frame = cap.read()
max_area = 0
frame = cv2.flip(frame, 3)
frame = cv2.resize(frame, (640, 480))
if ret == True:
if bkgrndSubMode == True:
roi = bkgrndSubMask(frame, x0, y0, width, height, framecount,
plot)
elif binaryMode == True:
roi = binaryMask(frame, x0, y0, width, height, framecount,
plot)
else:
roi = skinMask(frame, x0, y0, width, height, framecount, plot)
framecount = framecount + 1
end = time.time()
timediff = (end - start)
if (timediff >= 1):
#timediff = end - start
fps = 'FPS:%s' % (framecount)
start = time.time()
framecount = 0
cv2.putText(frame, fps, (10, 20), font, 0.7, (0, 255, 0), 2, 1)
cv2.putText(frame, 'Options:', (fx, fy), font, 0.7, (0, 255, 0), 2, 1)
cv2.putText(frame, 'b - Toggle Binary/SkinMask', (fx, fy + fh), font,
size, (0, 255, 0), 1, 1)
cv2.putText(frame, 'x - Toggle Background Sub Mask', (fx, fy + 2 * fh),
font, size, (0, 255, 0), 1, 1)
cv2.putText(frame, 'g - Toggle Prediction Mode', (fx, fy + 3 * fh),
font, size, (0, 255, 0), 1, 1)
cv2.putText(frame, 'q - Toggle Quiet Mode', (fx, fy + 4 * fh), font,
size, (0, 255, 0), 1, 1)
cv2.putText(frame, 'n - To enter name of new gesture folder',
(fx, fy + 5 * fh), font, size, (0, 255, 0), 1, 1)
cv2.putText(frame, 's - To start capturing new gestures for training',
(fx, fy + 6 * fh), font, size, (0, 255, 0), 1, 1)
cv2.putText(frame, 'ESC - Exit', (fx, fy + 7 * fh), font, size,
(0, 255, 0), 1, 1)
## If enabled will stop updating the main openCV windows
## Way to reduce some processing power :)
if not quietMode:
cv2.imshow('Original', frame)
cv2.imshow('ROI', roi)
if guessGesture == True:
#plot = np.zeros((512,512,3), np.uint8)
plot = myNN.update(plot, x)
x = (x + 1) % 2
cv2.imshow('Gesture Probability', plot)
#plot = np.zeros((512,512,3), np.uint8)
############## Keyboard inputs ##################
key = cv2.waitKey(5) & 0xff
## Use Esc key to close the program
if key == 27:
break
## Use b key to toggle between binary threshold or skinmask based filters
elif key == ord('b'):
binaryMode = not binaryMode
bkgrndSubMode = False
if binaryMode:
print("Binary Threshold filter active")
else:
print("SkinMask filter active")
## Use g key to start gesture predictions via CNN
elif key == ord('x'):
takebkgrndSubMask = True
bkgrndSubMode = True
print("BkgrndSubMask filter active")
## Use g key to start gesture predictions via CNN
elif key == ord('g'):
guessGesture = not guessGesture
print("Prediction Mode - {}".format(guessGesture))
## This option is not yet complete. So disabled for now
## Use v key to visualize layers
#elif key == ord('v'):
# visualize = True
## Use i,j,k,l to adjust ROI window
elif key == ord('i'):
y0 = y0 - 5
elif key == ord('k'):
y0 = y0 + 5
elif key == ord('j'):
x0 = x0 - 5
elif key == ord('l'):
x0 = x0 + 5
## Quiet mode to hide gesture window
elif key == ord('q'):
quietMode = not quietMode
print("Quiet Mode - {}".format(quietMode))
## Use s key to start/pause/resume taking snapshots
## numOfSamples controls number of snapshots to be taken PER gesture
elif key == ord('s'):
saveImg = not saveImg
if gestname != '':
saveImg = True
else:
print("Enter a gesture group name first, by pressing 'n'")
saveImg = False
## Use n key to enter gesture name
elif key == ord('n'):
gestname = input("Enter the gesture folder name: ")
try:
os.makedirs(gestname)
except OSError as e:
# if directory already present
if e.errno != 17:
print('Some issue while creating the directory named -' +
gestname)
path = "./" + gestname + "/"
#elif key != 255:
# print key
#Realse & destroy
cap.release()
cv2.destroyAllWindows()
cv2.waitKey(0)
def Main():
global guessGesture, visualize, mod, binaryMode, x0, y0, width, height, saveImg, gestname, path, img_flag, cap
font = cv2.FONT_HERSHEY_SIMPLEX
size = 0.7
fx = 10
fy = 355
fh = 18
#Call CNN model loading callback
while True:
# ans = int(raw_input(banner))
ans = 1
if ans == 2:
mod = myNN.loadCNN(-1)
myNN.trainModel(mod)
raw_input("Press any key to continue")
break
elif ans == 1:
print "Will load default weight file"
mod = myNN.loadCNN(0)
break
elif ans == 3:
if not mod:
w = int(raw_input("Which weight file to load (0 or 1)"))
mod = myNN.loadCNN(w)
else:
print "Will load default weight file"
img = int(raw_input("Image number "))
layer = int(raw_input("Enter which layer to visualize "))
myNN.visualizeLayers(mod, img, layer)
raw_input("Press any key to continue")
continue
else:
print "Get out of here!!!"
return 0
import os
print os.getpid()
## Grab camera input
cv2.namedWindow(
'Original',
cv2.WND_PROP_FULLSCREEN) # Set Camera Size, cv2.WINDOW_NORMAL
cv2.setWindowProperty('Original', cv2.WND_PROP_FULLSCREEN,
cv2.WINDOW_FULLSCREEN)
import signal
# MAC : SIGPROF 27
def signalCameraOnHandler(signum, frame):
print signum
global img_flag
global cap
cap = cv2.VideoCapture(0)
cv2.namedWindow('Original', cv2.WINDOW_NORMAL)
img_flag = "CAM"
# MAC : SIGINFO 29
def signalMapHandler(signum, frame):
print signum
global img_flag
global cap
cap = cv2.VideoCapture('map.gif')
cv2.namedWindow('Original', cv2.WINDOW_NORMAL)
img_flag = "MAP"
# MAC : SIGUSR1 30
def signalSmileFaceHandler(signum, frame):
print signum
global img_flag
global cap
cap = cv2.VideoCapture('smile_glow.gif')
cv2.namedWindow('Original', cv2.WINDOW_NORMAL)
img_flag = "SMILE"
# MAC : SIGUSR2 31
def signalDefaultFaceHandler(signum, frame):
print signum
global img_flag
global cap
cap = cv2.VideoCapture('normal_glow.gif')
cv2.namedWindow('Original', cv2.WINDOW_NORMAL)
img_flag = "NORMAL"
# Set Signal
signal.signal(signal.SIGPROF, signalCameraOnHandler)
signal.signal(signal.SIGINFO, signalMapHandler)
signal.signal(signal.SIGUSR1, signalSmileFaceHandler)
signal.signal(signal.SIGUSR2, signalDefaultFaceHandler)
reset_flag = False
thumbs_up_guess_stack = 0
smile_face_stack = 0
camera_page_stack = 0
map_page_stack = 0
while True:
ret, frame = cap.read()
# for end of gif
if not ret:
if img_flag == "NORMAL":
cap = cv2.VideoCapture('normal_glow.gif')
cv2.namedWindow('Original', cv2.WINDOW_NORMAL)
elif img_flag == "SMILE":
cap = cv2.VideoCapture('smile_glow.gif')
cv2.namedWindow('Original', cv2.WINDOW_NORMAL)
elif img_flag == "MAP":
cap = cv2.VideoCapture('map.gif')
cv2.namedWindow('Original', cv2.WINDOW_NORMAL)
ret, frame = cap.read()
if img_flag == "CAM":
camera_page_stack += 1
frame = cv2.flip(frame, 3)
guess_res = ":)"
if ret:
if binaryMode: # on
roi, guess_res = binaryMask(frame, x0, y0, width, height)
else:
roi, guess_res = skinMask(frame, x0, y0, width, height)
# detact thumbs-up frame
if guess_res.split()[0] == "PEACE" or guess_res.split()[0] == "OK":
camera_page_stack -= 1
thumbs_up_guess_stack += 1
else:
thumbs_up_guess_stack = 0
# print GUESS
cv2.putText(frame,
str(thumbs_up_guess_stack) + " " + guess_res, (fx, fy),
font, 1, (0, 0, 255), 2, 1)
cv2.putText(frame, "Show Thumbs-Up on TONY's hat camera!",
(fx, fy + fh), font, size, (0, 0, 255), 1, 1)
elif img_flag == "SMILE":
smile_face_stack += 1
cap = cv2.VideoCapture('smile_glow.gif')
elif img_flag == "NORMAL":
cap = cv2.VideoCapture('normal_glow.gif')
elif img_flag == "MAP":
map_page_stack += 1
cap = cv2.VideoCapture('map.gif')
# for signal call
if thumbs_up_guess_stack >= 6:
# paging to smile
os.system("kill -30 " + str(os.getpid()))
reset_flag = True
if smile_face_stack >= 30:
# paging to map
os.system("kill -29 " + str(os.getpid()))
reset_flag = True
if camera_page_stack >= 150:
# paging to map
os.system("kill -29 " + str(os.getpid()))
reset_flag = True
# now threshold == 300
if map_page_stack >= 200:
# paging to camera
os.system("kill -27 " + str(os.getpid()))
reset_flag = True
if reset_flag:
thumbs_up_guess_stack = 0
smile_face_stack = 0
camera_page_stack = 0
map_page_stack = 0
reset_flag = False
cv2.imshow('Original', frame)
key = cv2.waitKey(10) & 0xff
if key == 27:
break
#Realse & destroy
cap.release()
cv2.destroyAllWindows()
def Main():
# pdb.set_trace()
global guessGesture, visualize, mod, binaryMode, x0, y0, width, height, saveImg, gestname, path
quietMode = False
font = cv2.FONT_HERSHEY_SIMPLEX
size = 0.5
fx = 10
fy = 355
fh = 18
#Call CNN model loading callback
while True:
ans = int(raw_input( banner))
if ans == 2:
# 训练模型
mod = myNN.loadCNN(-1)
myNN.trainModel(mod)
raw_input("Press any key to continue")
break
elif ans == 1:
# 使用已有模型
print "Will load default weight file"
mod = myNN.loadCNN(0)
break
elif ans == 3:
# 打印某张手势图片经过每层的处理
if not mod:
w = int(raw_input("Which weight file to load (0 or 1)"))
mod = myNN.loadCNN(w)
else:
print "Will load default weight file"
img = int(raw_input("Image number "))
layer = int(raw_input("Enter which layer to visualize "))
myNN.visualizeLayers(mod, img, layer)
raw_input("Press any key to continue")
continue
else:
print "Get out of here!!!"
return 0
## Grab camera input
cap = cv2.VideoCapture(0)
cv2.namedWindow('Original', cv2.WINDOW_NORMAL)
# set rt size as 640x480
ret = cap.set(3,640)
ret = cap.set(4,480)
while(True):
ret, frame = cap.read()
max_area = 0
frame = cv2.flip(frame, 3)
if ret == True:
# ! 其间 可以做保存图片的操作
if binaryMode == True:
roi = binaryMask(frame, x0, y0, width, height)
else:
roi = skinMask(frame, x0, y0, width, height)
cv2.putText(frame,'Options:',(fx,fy), font, 0.7,(0,255,0),2,1)
cv2.putText(frame,'b - Toggle Binary/SkinMask',(fx,fy + fh), font, size,(0,255,0),1,1)
cv2.putText(frame,'g - Toggle Prediction Mode',(fx,fy + 2*fh), font, size,(0,255,0),1,1)
cv2.putText(frame,'q - Toggle Quiet Mode',(fx,fy + 3*fh), font, size,(0,255,0),1,1)
cv2.putText(frame,'n - To enter name of new gesture folder',(fx,fy + 4*fh), font, size,(0,255,0),1,1)
cv2.putText(frame,'s - To start capturing new gestures for training',(fx,fy + 5*fh), font, size,(0,255,0),1,1)
cv2.putText(frame,'ESC - Exit',(fx,fy + 6*fh), font, size,(0,255,0),1,1)
## If enabled will stop updating the main openCV windows
## Way to reduce some processing power :)
if not quietMode:
cv2.imshow('Original',frame) # 图像窗口
cv2.imshow('ROI', roi) # 截取手势窗口
# Keyboard inputs
key = cv2.waitKey(10) & 0xff
## Use Esc key to close the program
if key == 27:
break
## Use b key to toggle between binary threshold or skinmask based filters
elif key == ord('b'):
binaryMode = not binaryMode
if binaryMode:
print "Binary Threshold filter active"
else:
print "SkinMask filter active"
## Use g key to start gesture predictions via CNN
elif key == ord('g'):
guessGesture = not guessGesture
print "Prediction Mode - {}".format(guessGesture)
## This option is not yet complete. So disabled for now
## Use v key to visualize layers
# elif key == ord('v'):
# visualize = True
## Use i,j,k,l to adjust ROI window
elif key == ord('i'):
y0 = y0 - 5
elif key == ord('k'):
y0 = y0 + 5
elif key == ord('j'):
x0 = x0 - 5
elif key == ord('l'):
x0 = x0 + 5
## Quiet mode to hide gesture window
elif key == ord('q'):
quietMode = not quietMode
print "Quiet Mode - {}".format(quietMode)
## Use s key to start/pause/resume taking snapshots
## numOfSamples controls number of snapshots to be taken PER gesture
elif key == ord('s'):
saveImg = not saveImg
if gestname != '':
# 保存图片操作 binaryMask 或 skinMask 步骤操作
saveImg = True
else:
print "Enter a gesture group name first, by pressing 'n'"
saveImg = False
## Use n key to enter gesture name
elif key == ord('n'):
# 获取 手势名称 建立文件夹，然后会自动保存图片 到该目录
gestname = raw_input("Enter the gesture folder name: ")
try:
os.makedirs(gestname)
except OSError as e:
# if directory already present
if e.errno != 17:
print 'Some issue while creating the directory named -' + gestname
path = "./"+gestname+"/"
#elif key != 255:
# print key
#Realse & destroy
cap.release()
cv2.destroyAllWindows()
