def callback(data):
try:
img = bridge.imgmsg_to_cv2(data, "bgr8")
except CvBridgeError as e:
print(e)
time_start = clock()
img_detected = detect_face(img)
time_span = clock() - time_start
if time_span == 0:
fps = 0
else:
fps = 1 / time_span
draw_str(img_detected, (5, 30), 'fps: %d' % fps)
if show_video == True:
cv2.imshow('face detection', img_detected)
cv2.waitKey(1)
try:
pub.publish(bridge.cv2_to_imgmsg(img_detected, "bgr8"))
except CvBridgeError as e:
print(e)
def recog_openface(self, detection):
t = clock()
dinfo = []
for frame in detection.imgs:
if self.classif.equadra(frame.copy()):
#Classifica a imagem utilizando a rede neural do Openface
#Chama a funcao no arquivo Recog.py
#Retorna um rank de individuos possiveis, o primeiro tem maior probabilidade
resp, dinfo = self.classif.classifica()
(nome,pb,reffile,tipoc) = resp[0]
print "Encontrado {} em {} na camera {}".format(nome,detection.time, detection.camera.id)
#Cria um objeto com as informacoes sobre a deteccao
#Utilizado tambem para salvar a imagem e enviar para Kurento
#Grava a imagem original da deteccao
#d = Detection(detection.camera, nome, pb, detection.org_img, detection.time,send_alert)
#Grava a imagem cropada da deteccao
#d = Detection(detection.camera, nome, pb, frame.copy(), detection.time,True, dinfo)
#Grava a imagem original da deteccao
d = Detection(detection.camera, nome, pb, detection.org_img, detection.time,True, dinfo)
self.sync_queue_out.put(d)
dt = clock() - t
print 'T Time',dt,' s'
print "_____________________________________________________________________"
def callback(data):
bridge = CvBridge()
try:
cv_img = bridge.imgmsg_to_cv2(data, "bgr8")
except CvBridgeError as e:
print(e)
hog = cv2.HOGDescriptor()
hog.setSVMDetector(cv2.HOGDescriptor_getDefaultPeopleDetector())
time_start = clock()
detect_people(cv_img, hog)
time_span = clock() - time_start
if time_span == 0:
fps = 0
else:
fps = 1 / time_span
draw_str(cv_img, (5,30), 'fps: %d' % fps)
if show_video == True:
cv2.imshow('people detection', cv_img)
cv2.waitKey(1)
pub = rospy.Publisher("/opencv/detect/people",Image, queue_size = 1)
try:
pub.publish(bridge.cv2_to_imgmsg(cv_img, "bgr8"))
except CvBridgeError as e:
print(e)
def compare(face_to_check,learn=False):
import sys, getopt
detected_time = 0
detected_time_max = 10
video_src = 0
cascade_fn = os.path.join('data','haarcascades','haarcascade_frontalface_alt2.xml')
cascade = cv2.CascadeClassifier(cascade_fn)
cam = create_capture(video_src, fallback='synth:bg=../cpp/lena.jpg:noise=0.05')
while True:
ret, img1 = cam.read()
gray = cv2.cvtColor(img1, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects = detect(gray, cascade)
if len(rects):
if detected_time<detected_time_max:
detected_time+=1
else:
_found_size = (rects[0][0],rects[0][1],rects[0][2]-rects[0][0],
rects[0][3]-rects[0][1])
_found_face = cv.GetImage(cv.fromarray(img1))
cv.SetImageROI(_found_face,_found_size)
current_face = cv.CreateImage(cv.GetSize(_found_face),
_found_face.depth,
_found_face.nChannels)
if learn:
cv.Copy(_found_face, current_face, None)
cv.SaveImage(os.path.join('data','images',face_to_check),current_face)
cv.ResetImageROI(cv.GetImage(cv.fromarray(img1)))
img2 = cv.LoadImage(os.path.join('data','images',face_to_check))
dest_face = cv.CreateImage(cv.GetSize(img2),
img2.depth,
img2.nChannels)
cv.Resize(_found_face, dest_face)
if cv.Norm(dest_face,img2)<=30000:
return True
else:
return False
sys,exit()
else:
detected_time = 0
dt = clock() - t
def look_at_hog(img):
dims=(16,16)
t = clock()
hog = cv2.HOGDescriptor(dims,(16,16),(16,16),(8,8),9,1,-1)
feature_vector = hog.compute(img, winStride=dims, padding=(0,0))
dt = clock() - t
print 'Extraction took: %.1f ms' % (dt*1000)
t = clock()
hog_viz=draw_hog2(img,hog,feature_vector)
dt = clock() - t
print 'drawing took: %.1f ms' % (dt*1000)
return hog_viz
def callback(self, data):
cv_img=self.convert_image(data)
#cascade = cv2.CascadeClassifier("irobot_hog_detect.xml")
gray = cv2.cvtColor(cv_img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects = detect(gray, cascade)
vis = cv_img.copy()
draw_rects(vis, rects, (0, 255, 0))
dt = clock() - t
draw_str(vis, (20, 20), 'time: %.1f ms' % (dt*1000))
cv2.imshow('Irobot_Detector', vis)
def detect_and_draw(img, cascade):
# allocate temporary images
#gray = cv.CreateImage((img.width,img.height), 8, 1)
global haar_scale
global min_neighbors
for image_scale in range(1,5,1):
gray = cv.CreateImage((img.width,img.height), 8, 1)
small_img = cv.CreateImage((cv.Round(img.width / image_scale),cv.Round (img.height / image_scale)), 8, 1)
global noface
global current_time
global dailyfolder
current_time=time.strftime("%H:%M:%S")
dailyfolder=time.strftime("%F")
FileName="/Detected-Faces/"
FileName=dailyfolder+FileName+current_time+"_Image_Scale_"+str(image_scale)+ "_Min_Neighbors_" + str(min_neighbors) +".jpeg"
# convert color input image to grayscale
cv.CvtColor(img, gray, cv.CV_BGR2GRAY)
# scale input image for faster processing
cv.Resize(gray, small_img, cv.CV_INTER_LINEAR)
cv.EqualizeHist(small_img, small_img)
if(cascade):
t = clock()
faces = cv.HaarDetectObjects(small_img, cascade, cv.CreateMemStorage(0),haar_scale, min_neighbors, haar_flags, min_size)
t = clock() -t
if faces:
for ((x, y, w, h), n) in faces:
# the input to cv.HaarDetectObjects was resized, so scale the
# bounding box of each face and convert it to two CvPoints
pt1 = (int(x * image_scale), int(y * image_scale))
pt2 = (int((x + w) * image_scale), int((y + h) * image_scale))
cv.Rectangle(img, pt1, pt2, cv.RGB(255, 0, 0), 3, 8, 0)
#### DEBUG ###
#IO.LEDON()
#wait(10)
#IO.LEDOFF()
#save image update log
cv.SaveImage(FileName,img)
f.updatelog(t,image_scale,min_neighbors)
del(gray)
del(small_img)
else:
del(gray)
del(small_img)
def faceDetect(loopCount):
import sys, getopt
args, video_src = getopt.getopt(sys.argv[1:], "", ["cascade=", "nested-cascade="])
try:
video_src = video_src[0]
except:
video_src = 0
args = dict(args)
cascade_fn = args.get("--cascade", OpenCVInstallDir + "data/haarcascades/haarcascade_frontalface_alt.xml")
cascade = cv2.CascadeClassifier(cascade_fn)
cam = create_capture(video_src, fallback="synth:bg=../cpp/lena.jpg:noise=0.05")
idx = 0
if loopCount == 0:
loopCount = 1
infinteLoop = True
else:
infinteLoop = False
while idx < loopCount:
ret, img = cam.read()
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects = detect(gray, cascade)
vis = img.copy()
dt = clock() - t
draw_str(vis, (20, 20), "time: %.1f ms" % (dt * 1000))
if rects == []:
draw_str(vis, (20, 40), "We are having trouble seeing you, move around just a bit")
# draw_str(vis,(20,450), 'Look Here')
else:
if infinteLoop:
idx = 0
print rects
else:
idx = idx + 1
try:
rectsum = rectsum + rects
except:
rectsum = rects
# first time assignment
# cv2.imshow('facetracker', vis)
if 0xFF & cv2.waitKey(5) == 27:
break
cv2.destroyAllWindows()
return rectsum / idx
def get_face():
import sys, getopt
#print help_message
args, video_src = getopt.getopt(sys.argv[1:], '', ['cascade=', 'nested-cascade='])
try:
video_src = video_src[0]
except:
video_src = 0
args = dict(args)
cascade_fn = args.get('--cascade', "haarcascade_frontalface_alt.xml")
nested_fn = args.get('--nested-cascade', "haarcascade_eye.xml")
cascade = cv2.CascadeClassifier(cascade_fn)
nested = cv2.CascadeClassifier(nested_fn)
cam = create_capture(video_src, fallback='synth:bg=lena.jpg:noise=0.05')
just_face = ''
while True:
#pdb.set_trace()
ret, img = cam.read()
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects = detect(gray, cascade)
vis = img.copy()
# Change to crop image to only show face
if (just_face == ''):
just_face = img.copy()
param = (cv2.IMWRITE_PXM_BINARY, 1)
if (len(rects) > 0):
just_face = gray.copy()
(x1,y1,x2,y2) = rects[0]
just_face = just_face[y1:y2, x1:x2]
cv2.imwrite('./test_face.pgm', just_face, param)
return './test_face.pgm'
vis_roi = vis
draw_rects(vis, rects, (0, 255, 0))
dt = clock() - t
draw_str(vis, (20, 20), 'time: %.1f ms' % (dt*1000))
cv2.imshow('facedetect', just_face)
if 0xFF & cv2.waitKey(5) == 27:
break
cv2.destroyAllWindows()
def get_lbp_rects(img):
# print(img)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects = detect(gray)
dt = clock() - t
return rects
def on_frame(self, frame):
h, w = frame.shape[:2]
qi = 0
#print "on_frame %d x %d" % (h, w)
frame_diff = cv2.absdiff(frame, self.prev_frame)
gray_diff = cv2.cvtColor(frame_diff, cv2.COLOR_BGR2GRAY)
ret, motion_mask = cv2.threshold(gray_diff, self._threshold, 1, cv2.THRESH_BINARY)
timestamp = clock()
cv2.updateMotionHistory(motion_mask, self.motion_history, timestamp, MHI_DURATION)
mg_mask, mg_orient = cv2.calcMotionGradient(self.motion_history, MAX_TIME_DELTA, MIN_TIME_DELTA, apertureSize=5 )
seg_mask, seg_bounds = cv2.segmentMotion(self.motion_history, timestamp, MAX_TIME_DELTA)
centers = []
rects = []
draws = []
for i, rect in enumerate([(0, 0, w, h)] + list(seg_bounds)):
x, y, rw, rh = rect
area = rw*rh
if area < 64**2:
continue
silh_roi = motion_mask [y:y+rh,x:x+rw]
orient_roi = mg_orient [y:y+rh,x:x+rw]
mask_roi = mg_mask [y:y+rh,x:x+rw]
mhi_roi = self.motion_history[y:y+rh,x:x+rw]
if cv2.norm(silh_roi, cv2.NORM_L1) < area*0.05:
continue
angle = cv2.calcGlobalOrientation(orient_roi, mask_roi, mhi_roi, timestamp, MHI_DURATION)
color = ((255, 0, 0), (0, 0, 255))[i == 0]
if self._use_cv_gui:
draws.append(lambda vis, rect=rect, angle=angle, color=color:
draw_motion_comp(vis, rect, angle, color))
centers.append( (x+rw/2, y+rh/2) )
rects.append(rect)
self.tracker_group.update_trackers(centers, rects)
#print 'Active trackers: %d' % len(trackers)
#print 'Tracker score: %s' % ','.join(['%2d'%len(tracker.hits) for tracker in trackers])
trackers = self.tracker_group.trackers
cx, cy = None, None
#print "#trackers = %d" % len(trackers)
if len(trackers):
first_tracker = trackers[0]
cx, cy = center_after_median_threshold(frame, first_tracker.rect)
cv2.circle(frame, (cx, cy), 5, (255, 255, 255), 3)
print str(qi)*5; qi += 1
print self._on_cx_cy
self._on_cx_cy(cx, cy) # gives None's for no identified balloon
print str(qi)*5; qi += 1
if self._use_cv_gui:
self.on_frame_cv_gui(frame, draws, (cx, cy))
else:
self.frame_vis(frame, draws, (cx, cy))
#time.sleep(0.5)
self.prev_frame = frame.copy()
# TODO - print visualization onto image
return frame
def get_image(camera, headTracking):
global cascade
ret, im = camera.read()
t = clock()
if (headTracking):
smallIm = cv2.resize(im, (160,120))
grey = cv2.cvtColor(smallIm, cv2.COLOR_BGR2GRAY)
grey = cv2.equalizeHist(grey)
rects = detect(grey, cascade)
draw_rects(im, 4*rects, (0,255,0))
#TODO: Also fire servos we need.
draw_str(im, (20,40), str(readOutputPinsArduino()))
dt = clock() - t
draw_str(im, (20,20), 'Latency: %.4f ms' % (dt*1000))
im_rgb=cv2.cvtColor(im, cv2.COLOR_BGR2RGB)
cv_img=cv.fromarray(im_rgb)
return cv_img
def faceDetect(img,classifier_xml_dir):
cascade = cv2.CascadeClassifier(classifier_xml_dir)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects = detect(gray, cascade)
if len(rects)==0:
facesFound = 0
else:
facesFound = 1
vis = img.copy()
draw_rects(vis, rects, (0, 255, 0))
for x1, y1, x2, y2 in rects:
roi = gray[y1:y2, x1:x2]
vis_roi = vis[y1:y2, x1:x2]
dt = clock() - t
draw_str(vis, (20, 20), 'time: %.1f ms' % (dt*1000))
return (rects,facesFound)
def update(self):
# increment the total number of frames examined during the
# start and end intervals
self._numFrames += 1
if self._numFrames == self._window_size * 2:
self._numFrames -= 120
self._start = self._window_start
if self._numFrames == self._window_size:
self._window_start = clock()
def process_image(self):
"""
This function finds faces, draws them and their approximation by ellipses.
"""
t = clock()
rects = self.detect(self.gray, self.cascade)
vis = self.source_color
self.draw_rects(np.asarray(vis[:,:]), rects, (0, 255, 0))
for x1, y1, x2, y2 in rects:
roi = self.gray[y1:y2, x1:x2]
vis_roi = vis[y1:y2, x1:x2]
subrects = self.detect(roi.copy(), self.nested)
self.draw_rects(np.asarray(vis_roi[:,:]), subrects, (255, 0, 0))
dt = clock() - t
draw_str(np.asarray(vis[:,:]), (20, 20), 'time: %.1f ms' % (dt*1000))
cv2.imshow('facedetect', np.asarray(vis[:,:]))
cv2.waitKey(0)
cv2.destroyAllWindows()
def play_video(img, time_start, frame_count):
time_span = clock() - time_start
if time_span == 0:
fps = 0
fps = frame_count / time_span
draw_str(img, (5, 30), 'fps: %d' % fps)
cv2.imshow('play video', img)
if 0xFF & cv2.waitKey(1) == KEY_ECS:
raise ExitLoop
def init_scale(cam):
while True:
ret, img = cam.read()
draw_str(img, (20, 20), 'To initialize: Keep camera static and perpendicular to ground plane ')
draw_str(img, (20, 30), 'To initialize: Sit facing the camera approx. 2 feet away ')
draw_str(img, (20, 40), 'To initialize: Press <Space> to initialize ')
cv2.imshow('Initialize', img)
if 0xFF & cv2.waitKey(5) == 32:
break
cv2.destroyAllWindows()
print('Initializing')
nFaces = 0
avgHeight = 0
t = clock()
while True:
ret, img = cam.read()
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
rects = detect(gray, cascade)
draw_rects(img, rects, (0, 255, 0))
for x1, y1, x2, y2 in rects:
if( y2-y1 < 150 or y2-y1 > 220 ): # too small or too large; ignore as noise
continue
nFaces += 1
avgHeight = ((nFaces-1) * avgHeight + (y2-y1)) / nFaces # frontalface is square, so width is same
dt = clock() - t
draw_str(img, (20, 20), 'Initializing: Keep camera static and perpendicular to ground plane ')
draw_str(img, (20, 30), 'Initializing: Sit facing the camera approx. 2 feet away ')
if(dt <= 5.0):
draw_str(img, (20, imHeight - 20), 'time: %.1f ms' % (dt * 1000))
else:
draw_str(img, (20, imHeight - 20), 'time: DONE' )
cv2.imshow('Initializing', img)
if (0xFF & cv2.waitKey(5) == 32) or dt > 5.0:
break
cv2.destroyAllWindows()
return avgHeight
def __init__(self, topic, should_mirror, verbose):
self.topic = topic
self.KEY_ECS = 27
self.should_mirror = should_mirror
self.verbose = verbose
self.bridge = CvBridge()
rospy.on_shutdown(self.cleanup)
self.shutdowm_msg = "Shutting down."
self.node_name = "image_subscriber"
self.time_start = clock()
self.fps = FPS()
def showImg(self, frame, keypoints, lines, contours):
if self.args.nodisplay and self.args.stashinterval == 0:
return
if keypoints:
frame = cv2.drawKeypoints(
frame, [keypoints[0]], np.array([]), (0, 0, 255), cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS
)
if len(keypoints) > 1:
frame = cv2.drawKeypoints(
frame, keypoints[1:], np.array([]), (255, 205, 25), cv2.DRAW_MATCHES_FLAGS_DRAW_RICH_KEYPOINTS
)
if lines != None:
for l in lines[0]:
cv2.line(frame, (l[0], l[1]), (l[2], l[3]), (20, 255, 255))
if contours != None:
contours0, hier = contours
cindex = self.values[3] # if -1, all are drawn
maxlevel = self.values[4]
if len(contours0) <= cindex:
self.putNotice("reset contour id")
values[3] = -1
cindex = -1
cv2.drawContours(
frame,
contours0,
cindex,
(128, 255, 255),
thickness=1,
lineType=cv2.CV_AA,
hierarchy=hier,
maxLevel=maxlevel,
)
if not self.args.nodisplay:
cv2.imshow("img", frame)
if self.args.stashinterval != 0 and (common.clock() - self.lastStashTime) > self.args.stashinterval:
cv2.imwrite(self.stashFilename, frame, self.stashParams)
self.lastStashTime = common.clock()
def main():
import sys, getopt
args, video_src = getopt.getopt(sys.argv[1:], '', ['cascade=', 'nested-cascade='])
try:
video_src = video_src[0]
except:
video_src = 0
args = dict(args)
cascade_fn = args.get('--cascade', "data/haarcascades/haarcascade_frontalface_alt.xml")
nested_fn = args.get('--nested-cascade', "data/haarcascades/haarcascade_eye.xml")
cascade = cv.CascadeClassifier(cv.samples.findFile(cascade_fn))
nested = cv.CascadeClassifier(cv.samples.findFile(nested_fn))
cam = create_capture(video_src, fallback='synth:bg={}:noise=0.05'.format(cv.samples.findFile('samples/data/lena.jpg')))
while True:
ret, img = cam.read()
gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY)
gray = cv.equalizeHist(gray)
t = clock()
rects = detect(gray, cascade)
vis = img.copy()
draw_rects(vis, rects, (0, 255, 0))
if not nested.empty():
for x1, y1, x2, y2 in rects:
roi = gray[y1:y2, x1:x2]
vis_roi = vis[y1:y2, x1:x2]
subrects = detect(roi.copy(), nested)
draw_rects(vis_roi, subrects, (255, 0, 0))
dt = clock() - t
draw_str(vis, (20, 20), 'time: %.1f ms' % (dt*1000))
cv.imshow('facedetect', vis)
if cv.waitKey(5) == 27:
break
print('Done')
def show_video(self, img):
self.fps.update()
draw_str(img, (5, 30), "fps: %s" % self.fps)
cv2.imshow("show %s" % (self.topic), img)
key = cv2.waitKey(1)
if 0xFF & key == self.KEY_ECS:
rospy.signal_shutdown("User hit q key to quit.")
elif 0xFF & key == ord("a"):
file_name = "image_%s.jpg" % (str(int(clock())))
cv2.imwrite(file_name, img)
print "%s has saved." % file_name
def __init__(self, topic, should_mirror, verbose):
self.topic = topic
self.KEY_ECS = 27
self.should_mirror = should_mirror
self.verbose = verbose
self.bridge = CvBridge()
rospy.on_shutdown(self.cleanup)
self.shutdowm_msg = "Shutting down."
self.node_name = 'image_subscriber'
self.time_start = clock()
self.frames = []
self.frame_max = 90
self.imgPlayer = ImagePlayer("show %s" % (self.topic))
def __init__(self, stream):
self.stream = stream
self.numThread = cv2.getNumberOfCPUs()
#self.numThread = 1
self.workerPool = ThreadPool(processes = self.numThread)
self.pendingWorker = deque()
self.latency = StatValue()
self.frameInterval = StatValue()
self.lastFrameTime = clock()
self.outFrames = deque(maxlen = self.numThread)
self.faces = []
def show_video(self):
while not rospy.is_shutdown():
if len(self.frames) > 0:
img = self.frames.pop(0)
self.imgPlayer.show(img)
key = self.imgPlayer.get_key()
if key == self.KEY_ECS:
rospy.signal_shutdown("User hit q key to quit.")
elif key == ord('a'):
file_name = 'image_%s.jpg' % (str(int(clock())))
cv2.imwrite(file_name, img)
print '%s has saved.' % file_name
def threadedProcess(self):
rects = []
if len(self.pendingWorker) > 0 and self.pendingWorker[0].ready():
task = self.pendingWorker.popleft()
frame, curTime = task.get()
self.latency.update(clock() - curTime)
draw_str(frame, (20, config.VIDEO_WIDTH -20), "Latency: %.1f ms" % (self.latency.value*1000))
draw_str(frame, (20, config.VIDEO_WIDTH - 35), "FPS: %d" % (1/self.frameInterval.value))
#print("Latency %lf" % (self.latency.value*1000))
#print("FPS: %d" % (1/self.frameInterval.value))
self.outFrames.append(frame)
#cv2.imshow('Processed Video', frame)
#cv2.waitKey(1)
if len(self.pendingWorker) < self.numThread:
frame = self.stream.read()
t = clock()
self.frameInterval.update(t - self.lastFrameTime)
self.lastFrameTime = t
task = self.workerPool.apply_async(process, (copy.copy(frame), t))
self.pendingWorker.append(task)
def threadedProcess(self):
rects = []
if len(self.pendingWorker) > 0 and self.pendingWorker[0].ready():
task = self.pendingWorker.popleft()
frame, curTime = task.get()
self.latency.update(clock() - curTime)
draw_str(frame, (20, 360-20), "Latency: %.1f ms" % (self.latency.value*1000))
draw_str(frame, (20, 360- 35), "FPS: %d" % (1/self.frameInterval.value))
self.outFrames.append(frame)
'''
if len(self.pendingWorker) > 0:
for i in range(0, len(self.pendingWorker)):
if self.pendingWorker[i].ready():
for j in range(0, i):
waste = self.pendingWorker.popleft()
try:
waste.get()
except:
pass
task = self.pendingWorker.popleft()
frame, time = task.get()
self.latency.update(clock() - time)
draw_str(frame, (20, 20), "Latency: %.1f ms" % (self.latency.value*1000))
draw_str(frame, (300, 20), "FPS: %d" % (1/self.frameInterval.value))
cv2.imshow('Processed Video', frame)
cv2.waitKey(1)
break
'''
if len(self.pendingWorker) < self.numThread:
grab, frame = self.stream.read()
t = clock()
self.frameInterval.update(t - self.lastFrameTime)
self.lastFrameTime = t
task = self.workerPool.apply_async(process, (copy.copy(frame), t))
self.pendingWorker.append(task)
def handle_pub(video_path):
topic = '/file/video'
rospy.init_node('video_publisher')
pub = rospy.Publisher(topic, Image, queue_size=2)
print "\npublish video to topic:%s from file:%s ..." % (topic, video_path)
videoCapture = cv2.VideoCapture(video_path)
bridge = CvBridge()
rate = rospy.Rate(FPS)
time_start = clock()
frame_count = 0
success, img = videoCapture.read()
while success:
img_copy = img.copy()
try:
msg = bridge.cv2_to_imgmsg(img, "bgr8")
except CvBridgeError as e:
print(e)
pub.publish(msg)
if show_video == True:
time_span = clock() - time_start
if time_span == 0:
fps = 0
fps = frame_count / time_span
draw_str(img_copy, (5, 30), 'fps: %d' % fps)
cv2.imshow('play video', img_copy)
if 0xFF & cv2.waitKey(1) == KEY_ECS:
break
rate.sleep()
success, img = videoCapture.read()
frame_count += 1
cv2.destroyAllWindows()
def capture(self):
string = ""
ret, img = self.cam.read()
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects = self.detect(gray, self.cascade)
if len(rects) == 0:
string = str(int(round(time.time() * 1000))) + ";n.f.;"
vis = img.copy()
self.draw_rects(vis, rects, (0, 255, 0))
for x1, y1, x2, y2 in rects:
roi = gray[y1:y2, x1:x2]
vis_roi = vis[y1:y2, x1:x2]
subrects_fn = self.detect(roi.copy(), self.nested)
subrects_glass = self.detect(roi.copy(), self.glass)
subrects_le = self.detect(roi.copy(), self.le)
subrects_re = self.detect(roi.copy(), self.re)
string = string + str(int(round(time.time() * 1000))) + ";"
if not len(subrects_fn) == 0:
self.draw_rects(vis_roi, subrects_fn, (255, 0, 0))
string = string + "1;"
elif not len(subrects_glass) == 0:
self.draw_rects(vis_roi, subrects_glass, (255, 0, 0))
string = string + "1;"
elif (not len(subrects_le) == 0) or (not len(subrects_re) == 0):
self.draw_rects(vis_roi, subrects_le, (255, 0, 0))
self.draw_rects(vis_roi, subrects_re, (255, 0, 0))
string = string + "0;"
else:
string = string + "n.e.;"
dt = clock() - t
draw_str(vis, (20, 20), 'time: %.1f ms' % (dt*1000))
cv2.imshow('facedetect', vis)
cv2.imwrite(self.foldername + "/eyeCaptureImages/" + str(int(round(time.time()*1000))) + ".jpg", vis)
return string
def __init__(self, topic, should_mirror, verbose, limit=1000):
self.topic = topic
self.KEY_ECS = 27
self.should_mirror = should_mirror
self.verbose = verbose
self.bridge = CvBridge()
rospy.on_shutdown(self.cleanup)
self.shutdowm_msg = "Shutting down."
self.node_name = 'image_subscriber'
self.time_start = clock()
self.limit = limit
self.frame_count = 0
self.total_latency = 0
self.fps = FPS()
def update(self, frame):
frame_diff = cv2.absdiff(frame, self.prev_frame)
real_diff = frame - self.prev_frame
self.real_diff = cv2.cvtColor(real_diff, cv2.COLOR_BGR2GRAY)
gray_diff = cv2.cvtColor(frame_diff, cv2.COLOR_BGR2GRAY)
thrs = 40 #cv2.getTrackbarPos('threshold', 'motempl')
ret, motion_mask = cv2.threshold(gray_diff, thrs, 1, cv2.THRESH_BINARY)
timestamp = clock()
cv2.updateMotionHistory(motion_mask, self.motion_history, timestamp, MHI_DURATION)
self.vis = np.uint8(np.clip((self.motion_history-(timestamp-MHI_DURATION)) / MHI_DURATION, 0, 1)*255)
self.vis = cv2.cvtColor(self.vis, cv2.COLOR_GRAY2BGR)
#self.process_motions()
self.prev_frame = frame.copy()
else:
if not platform.system() == 'Windows':
print("Unsupported format in capture check ", fourcc)
break
else:
imglR = cv2.resize(img, (640, 360))
gray = cv2.cvtColor(imglR, cv2.COLOR_BGR2GRAY)
if show_params:
cv2.putText(img, "Mode: {}".format(fourcc), (15, 40), font, 1.0,
color)
cv2.putText(img, "FPS: {}".format(fps), (15, 80), font, 1.0, color)
vis = img.copy()
if (fotograma % detectar) == 0:
t = clock()
rects = detect(gray, cascade)
#draw_rects(vis, rects, (0, 255, 0))
rects2 = detect(gray, cascade2)
#draw_rects(vis, rects2, (255, 255, 0))
ret = ru.promediarDetecciones(rects, rects2)
if show_params:
draw_rects(vis, np.array(ret, dtype=np.int32), (0, 255, 0),
scale)
dt = clock() - t
if show_params:
draw_str(vis, (20, 20),
'time Detection: %.1f ms' % (dt * 1000))
draw_str(vis, (200, 20),
'time Tracking: %.1f ms' % (dt2 * 1000))
boxes = multiTracker.update(imglR)
def Run(self):
if self.args.nodisplay:
print("\nimgExplore: running in nodisplay mode (keys inactive)")
else:
print("\n\nkeys:\n" \
" ESC: exit\n\n" \
" c: rgb\n" \
" r: red\n" \
" g: green\n" \
" b: blue,\n" \
" h: hue\n" \
" s: sat\n"\
" v: val\n" \
" 1: adaptive threshold (thresh,inv)\n"\
" 2: threshold (thresh,inv)\n"\
" 3: huerange*valrange (hmin,hmax,vmin,vmax)\n"\
" 4: canny edges (thresh/10, range)\n"\
" 5: simple blobs (thresh0,thresh1,treshStep,minA,colorthresh)\n"\
" 6: houghlines (rho,theta,thresh,minlen,maxgap)\n"\
" 7: contours (mode:0-3,method:0-3,offset,id(-1:all),depth)\n"\
" 8: ORB features (nfeatures,scaleFactor(0->255)],patchSize)\n"\
" 9: dance1 (minX,maxX)\n"\
" 0: gamma (gamma)\n"\
"\n"\
" F1,F2: -/+ v1\n"\
" F3,F4: -/+ v2\n"\
" F5,F6: -/+ v3\n"\
" F7,F8: -/+ v4\n"\
" F9,F10: -/+ v5\n"\
" F11,F12: -/+ v6\n"\
"\n"\
" <right arrow>: increase img seq frame\n"\
" <left arrow>: decrease img seq frame\n"\
)
if not self.fnpat:
for i in range(0, 4):
vsrc = cv2.VideoCapture(i)
if not vsrc or not vsrc.isOpened():
print("Problem opening video source %d" % i)
vsrc = None
else:
break
if not vsrc:
exit(1)
else:
ret1 = vsrc.set(cv2.cv.CV_CAP_PROP_FRAME_WIDTH, 1280)
ret2 = vsrc.set(cv2.cv.CV_CAP_PROP_FRAME_HEIGHT, 720)
print(ret1, ret2)
if 1:
w = vsrc.get(cv2.cv.CV_CAP_PROP_FRAME_WIDTH)
h = vsrc.get(cv2.cv.CV_CAP_PROP_FRAME_HEIGHT)
print("video res: %d %d" % (w, h))
else:
vsrc = None
self.update = False
while True:
while len(self.pending) > 0 and self.pending[0].ready():
frame, t0, keypoints, lines, contours = self.pending.popleft(
).get()
self.latency.update(common.clock() - t0)
self.robotCnx.SetFPS(int(1 / self.frameT.value))
self.drawStr(
frame, (20, 20),
"latency : %.1f ms" % (self.latency.value * 1000))
self.drawStr(
frame, (20, 40),
"frame interval: %.1f ms" % (self.frameT.value * 1000))
self.showImg(frame, keypoints, lines, contours)
if vsrc:
# Here we have a video source... Capture the image in the
# main thread, process it in other threads.
if len(self.pending) < self.threadn:
# we have threads available to perform work for us
ret, self.mainImg = vsrc.read() # <---can take some time
t = common.clock()
self.frameT.update(t - self.lastFrameTime)
self.lastFrameTime = t
# pass a copy of the new frame to another thread for
# processing this alows us to get back to the
# time-consuming task of capturing the next video frame.
task = self.pool.apply_async(
processFrameCB, (self, self.mainImg.copy(), t))
self.pending.append(task)
done,self.update,self.cmode,self.index,self.values,msg = \
self.checkKey(1, self.cmode, self.index, self.values)
if msg:
self.putStatus(msg)
if done:
break
else:
# Here we don't have a video source, rather rely on 'canned'
# images. Thus, we don't use multiple threads... just do it
# all right here.
if self.index < 0:
# index < 0 signals user's intent to go bkd in framelist
goback = True
self.index = abs(self.index)
else:
goback = False
fn = self.fnpat % self.index
base = os.path.basename(fn)
if not os.path.isfile(fn) and self.index >= 0:
# image sequence may be missing some frames..
self.putNotice(base + " not found")
if goback:
self.index = -(self.index - 1)
else:
self.index = self.index + 1
continue
if fn != self.lastFn or self.update:
if not self.update:
# update the current filename status field
self.putNotice(base)
self.mainImg = cv2.imread(fn, cv2.IMREAD_ANYCOLOR)
(img, t0, keypts, lines,
contours) = self.processFrame(self.mainImg.copy(),
common.clock())
self.frameT.update(common.clock() - t0)
self.robotCnx.SetFPS(int(1 / self.frameT.value))
if not self.args.nodisplay:
if self.cmode == 'rgb':
str = "%s (%.2f ms)" % \
(base,self.frameT.value*1000)
else:
str = "%s[%s] (%.2f ms)" % \
(base,self.cmode,self.frameT.value*1000)
self.drawStr(img, (20, 20), str)
self.showImg(img, keypts, lines, contours)
self.lastFn = fn
done,self.update,self.cmode,self.index,self.values,msg = \
self.checkKey(10, self.cmode, self.index, self.values)
if msg:
self.putStatus(msg)
if done:
break
# end of while
self.robotCnx.Shutdown()
def process_frame(frame, t0):
# some intensive computation...
frame = cv2.medianBlur(frame, 19)
# frame = cv2.medianBlur(frame, 19)
return frame, t0
threadn = cv2.getNumberOfCPUs()
pool = ThreadPool(processes = threadn)
pending = deque()
threaded_mode = True
latency = StatValue()
frame_interval = StatValue()
last_frame_time = clock()
while True:
while len(pending) > 0 and pending[0].ready():
''' '''
res, t0 = pending.popleft().get()
latency.update(clock() - t0)
draw_str(res, (20, 20), "Latency: %.1f ms" % (latency.value*1000))
draw_str(res, (100, 20), "Frame interval: %.1f ms" % (frame_interval.value*1000))
print('Interval: %.lf ms',(frame_interval.value*1000))
#cv2.imshow('threaded video', res)
frame = cv2.medianBlur(frame, 19)
if len(pending) < threadn:
#camera.capture(rawCap, format = "bgr")
#frame = rawCap.array
frame = cap.read()
t = clock()
def main(cam_id):
"""
由于眼部检测的不稳定性，这里采用在已经识别出的人脸范围内假定的眼睛区域作为眼睛区域的判断
采用 关键点检测 加 伪眼部光流检测
关键点静止但眼部光流明显时判定为活体
这个主要判断 眨眼 来做为活体检测的依据
:return:
"""
print(__doc__)
cap = cv2.VideoCapture(cam_id)
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('D:/data.avi', fourcc, 30, (640, 480))
f_cascade = cv2.CascadeClassifier(
"C:/opencv/opencv/build/etc/haarcascades/haarcascade_frontalface_alt2.xml"
)
e_cascade = cv2.CascadeClassifier(
"C:\\opencv\\opencv\\build\etc\\haarcascades\\haarcascade_eye.xml")
ret, prev = cap.read()
prev_gray = cv2.cvtColor(prev, cv2.COLOR_BGR2GRAY)
lk_params = dict(winSize=(15, 15),
maxLevel=2,
criteria=(cv2.TERM_CRITERIA_EPS | cv2.TERM_CRITERIA_COUNT,
10, 0.03))
feature_params = dict(maxCorners=500,
qualityLevel=0.3,
minDistance=7,
blockSize=7)
tracks = []
frame_index = 0
detect_interval = 3
track_len = 10
msg_show_opt = 0 # 通过信息显示帧数
msg_show_key = 0 # 通过信息显示帧数
has_face = False
# 存储每一帧的光流
eye_flow_lines_t = []
while True:
if cv2.waitKey(1) == 27: # Esc for exit
break
t = clock()
ret, img = cap.read()
img = cv2.flip(img, 1)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
rectangles = detect(gray, f_cascade)
mask = np.zeros_like(gray) # 设置关键点遮罩
if len(rectangles) == 1: # 限制一张人脸
if not (has_face and True):
tracks = []
has_face = True
for rectangle in rectangles:
rx0, ry0, rx1, ry1 = rectangle
# if not (140 < rx1 - rx0 < 160 and 140 < ry1 - ry0 < 160): # 限定人脸识别框的大小
# continue
draw_rectangle(img, rectangle, color=(0, 225, 0)) # 人脸范围
rectangles_eye = detect(gray[ry0:ry1, rx0:rx1],
e_cascade) # 获取眼睛范围
# draw_rectangles(img[ry0:ry1, rx0:rx1], rectangles_eye, color=(255, 0, 225))
# 眼部光流场功能
eye_flow_lines = []
# for erx0, ery0, erx1, ery1 in rectangles_eye:
# eye_flow = opt_flow(prev_gray[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1],
# gray[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1]) # get opt flow
# eye_flow_lines.append(draw_flow(img[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1],
# eye_flow, step=4)) # 显示光流点
# 假眼部位置，假设脸纵向上部1/4位置到2/4位置及 横向左部1/6到5/6位置为眼部，以抵消眼部识别不能每次都有效地问题
face_h = ry1 - ry0
face_w = rx1 - rx0
face_hs = face_h / 4
face_he = face_h / 2
face_ws = face_w / 6
face_we = face_w / 6 * 5
eye_flow = opt_flow(
prev_gray[ry0:ry1, rx0:rx1][face_hs:face_he,
face_ws:face_we],
gray[ry0:ry1, rx0:rx1][face_hs:face_he, face_ws:face_we])
eye_flow_lines.append(
draw_flow(img[ry0:ry1, rx0:rx1][face_hs:face_he,
face_ws:face_we],
eye_flow,
step=4))
eye_sorted = [] # 排序后的长度集合(眼睛)
eye_sorted2 = []
for lines in eye_flow_lines:
mds = []
for (x1, y1), (x2, y2) in lines:
md = math.sqrt((x2 - x1)**2 + (y2 - y1)**2)
mds.append(md)
eye_sorted2.append(md)
eye_sorted.append(sorted(mds, reverse=True))
eye_flow_lines_t.append(eye_sorted2) # 存储每一帧的光流位移信息
# 绘制关键点轨迹
# 会删除位移较大的关键点
# 不影响其他的鉴别
if len(tracks) > 0:
img0, img1 = prev_gray, gray
p0 = np.float32([tr[-1]
for tr in tracks]).reshape(-1, 1, 2)
p1, st, err = cv2.calcOpticalFlowPyrLK(
img0, img1, p0, None, **lk_params)
p0r, st, err = cv2.calcOpticalFlowPyrLK(
img1, img0, p1, None, **lk_params)
d = abs(p0 - p0r).reshape(-1, 2).max(-1)
good = d < 0.5
new_tracks = []
for tr, (x, y), good_flag in zip(tracks, p1.reshape(-1, 2),
good):
if not good_flag:
continue
if not (rx0 < x < rx1 and ry0 < y < ry1):
continue
tr.append((x, y))
if len(tr) > track_len:
del tr[0]
new_tracks.append(tr)
cv2.circle(img, (x, y), 2, (0, 255, 0), -1)
tracks = new_tracks
cv2.polylines(img, [np.int32(tr) for tr in tracks], False,
(0, 255, 0))
draw_str(img, (20, 20), 'track count: %d' % len(tracks))
# 限定人脸为兴趣区域
cv2.fillPoly(
mask,
np.array([[[rx0, ry0], [rx1, ry0], [rx1, ry1],
[rx0, ry1]]]), (255, 255, 255))
for x, y in [np.int32(tr[-1]) for tr in tracks]:
cv2.circle(mask, (x, y), 5, 0, -1) # 排除上一次的关键点
if frame_index % detect_interval == 0:
print('**************** start ***************')
l_sorted = []
l_sorted_eye = [] # 眼睛区域的关键点
l_sorted_out = [] # 眼睛外部的关键点
l_tmp = []
l_tmp_eye = []
l_tmp_out = []
for tr in tracks:
(x0, y0) = tr[0]
(x1, y1) = tr[-1]
if rx0 + face_ws < x1 < rx0 + face_we and ry0 + face_hs < y1 < ry1 + face_he:
l_tmp_eye.append(
round(math.sqrt((x1 - x0)**2 + (y1 - y0)**2),
2))
else:
l_tmp_out.append(
round(math.sqrt((x1 - x0)**2 + (y1 - y0)**2),
2))
l = round(math.sqrt((x1 - x0)**2 + (y1 - y0)**2), 2)
l_tmp.append(l)
# if l > 0:
# print(round(math.atan(abs((y1 - y0) / (x1 - x0))) / math.pi * 180, 2), end=':')
print(l, end='\t')
print('\n+++++++++++++++')
l_sorted = sorted(l_tmp, reverse=True)
l_sorted_eye = sorted(l_tmp_eye, reverse=True)
l_sorted_out = sorted(l_tmp_out, reverse=True)
if len(l_sorted_out) > 3 and len(l_sorted_eye) > 3 \
and l_sorted_out[0] < 1 and l_sorted_eye[0] > 1 \
and l_sorted_eye[0] - l_sorted_out[0] > 1:
print('xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx')
msg_show_key = 30
# ========打印前十个=========================
if True:
for i, md2 in enumerate(eye_sorted):
count = 0
print('眼睛', str(i + 1), end=':\t')
for md in md2:
count += 1
if count > 150:
break
print(round(md, 2), end=',')
print()
print('###################')
# 活体检测
np_eye = np.array(sorted(eye_sorted2, reverse=True)[:30])
np_eye = np_eye[np_eye > 0]
np_l = np.array(l_sorted[:10])
print(np_eye.size, '+++++', np_l.size)
if np_eye.size != 0 and np_l.size != 0:
flow_pre = np_eye[np_eye > 2].size * 1.0 / np_eye.size
ln_pre = np_l[np_l > 2].size * 1.0 / np_l.size
print(flow_pre, '---', ln_pre)
if 0.8 > flow_pre > 0.05 and ln_pre < 0.2:
msg_show_opt = 30
print(
'yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy'
)
print('**************** end ***************')
# 判断关键点
p = cv2.goodFeaturesToTrack(gray,
mask=mask,
**feature_params)
if p is not None:
for x, y in np.float32(p).reshape(-1, 2):
tracks.append([(x, y)])
frame_index += 1
else:
has_face = False
prev_gray = gray
dt = clock() - t
# draw_str(img, (20, 20), 'time: %.1f ms' % (dt * 1000))
if msg_show_key > 0:
draw_str(img, (450, 20), 'YES by KEY', front=(0, 0, 255))
msg_show_key -= 1
if msg_show_opt > 0:
draw_str(img, (300, 20), 'YES by OPT', front=(0, 0, 255))
msg_show_opt -= 1
cv2.imshow("Face detect" + str(cam_id), img)
out.write(img)
# cv2.imshow('mask', mask)
cap.release()
out.release()
cv2.destroyAllWindows()
cam = video.create_capture(
video_src, fallback='synth:class=chess:bg=../cpp/lena.jpg:noise=0.01')
ret, frame = cam.read()
h, w = frame.shape[:2]
prev_frame = frame.copy()
motion_history = np.zeros((h, w), np.float32)
hsv = np.zeros((h, w, 3), np.uint8)
hsv[:, :, 1] = 255
while True:
ret, frame = cam.read()
frame_diff = cv2.absdiff(frame, prev_frame)
gray_diff = cv2.cvtColor(frame_diff, cv2.COLOR_BGR2GRAY)
thrs = cv2.getTrackbarPos('threshold', 'motempl')
ret, motion_mask = cv2.threshold(gray_diff, thrs, 1, cv2.THRESH_BINARY)
print type(motion_mask[0][0])
timestamp = clock()
cv2.motempl.updateMotionHistory(motion_mask, motion_history, timestamp,
MHI_DURATION)
mg_mask, mg_orient = cv2.motempl.calcMotionGradient(motion_history,
MAX_TIME_DELTA,
MIN_TIME_DELTA,
apertureSize=5)
seg_mask, seg_bounds = cv2.motempl.segmentMotion(
motion_history, timestamp, MAX_TIME_DELTA)
visual_name = visuals[cv2.getTrackbarPos('visual', 'motempl')]
if visual_name == 'input':
vis = frame.copy()
elif visual_name == 'frame_diff':
vis = frame_diff.copy()
elif visual_name == 'motion_hist':
def image_cap():
d=[]
print(__doc__)
try:
fn = sys.argv[1]
except:
fn = 0
cap = video.create_capture(fn)
cap.set(cv.CAP_PROP_FPS, 12)
def process_frame(frame, t0):
# some intensive computation...
#frame = cv.medianBlur(frame, 19)
#frame = cv.medianBlur(frame, 19)
return frame, t0
threadn = cv.getNumberOfCPUs()
pool = ThreadPool(processes = threadn)
pending = deque()
threaded_mode = True
latency = StatValue()
frame_interval = StatValue()
last_frame_time = clock()
while True:
while len(pending) > 0 and pending[0].ready():
res, t0 = pending.popleft().get()
latency.update(clock() - t0)
draw_str(res, (20, 20), "threaded : " + str(threaded_mode))
draw_str(res, (20, 40), "latency : %.1f ms" % (latency.value*1000))
draw_str(res, (20, 60), "frame interval : %.1f ms" % (frame_interval.value*1000))
res=cv.resize(res,(176,100))
cv2_im = cv.cvtColor(res, cv.COLOR_BGR2RGB)
pil_im = Image.fromarray(cv2_im)
d.append(np.array(pil_im))
if len(d)==32:
t1 = data_input(d[0:16])
t2 = data_input(d[16:32])
in_x = np.array([t1, t2])
in_x = np.reshape(in_x, (2, 16, 128, 128, 3))
start = time.clock()
#p = Pool(1)
#p.map(evaluate, in_x)
evaluate(in_x)
elapsed = time.clock()
elapsed = elapsed - start
print("Time spent in (function name) is: ", elapsed)
d=[]
cv.imshow('threaded video', res)
if len(pending) < threadn:
ret, frame = cap.read()
t = clock()
frame_interval.update(t - last_frame_time)
last_frame_time = t
if threaded_mode:
task = pool.apply_async(process_frame, (frame.copy(), t))
else:
task = DummyTask(process_frame(frame, t))
pending.append(task)
ch = cv.waitKey(1)
if ch == ord(' '):
threaded_mode = not threaded_mode
if ch == 27:
break
cv.destroyAllWindows()
known_face_names.append(re.sub("[0-9]", '', filename[:-4]))
known_face_encodings.append(face_recognition.face_encodings(face)[0])
face_locations = []
face_encodings = []
face_names = []
process_this_frame = True
# get pluged camera, otherwise, default image.
cam = create_capture(0,
fallback='synth:bg={}:noise=0.05'.format(
cv.samples.findFile('src/image/image.jpg')))
while True:
_, frame = cam.read()
tempsAvantCalculs = clock()
if process_this_frame:
face_locations = face_recognition.face_locations(frame)
face_encodings = face_recognition.face_encodings(frame, face_locations)
face_names = []
for face_encoding in face_encodings:
matches = face_recognition.compare_faces(known_face_encodings,
face_encoding)
name = "Inconnu"
face_distances = face_recognition.face_distance(
known_face_encodings, face_encoding)
best_match_index = np.argmin(face_distances)
def main():
"""
选取八个区域进行规律查找
:return:
"""
print __doc__
cap = cv2.VideoCapture(0)
f_cascade = cv2.CascadeClassifier("C:/opencv/opencv/build/etc/haarcascades/haarcascade_frontalface_alt.xml")
e_cascade = cv2.CascadeClassifier("C:\\opencv\\opencv\\build\etc\\haarcascades\\haarcascade_eye.xml")
ret, prev = cap.read()
prev_gray = cv2.cvtColor(prev, cv2.COLOR_BGR2GRAY)
track_len = 10
detect_interval = 1
tracks = []
frame_idx = 0
lk_params = dict(winSize=(15, 15), maxLevel=2, criteria=(cv2.TERM_CRITERIA_EPS | cv2.TERM_CRITERIA_COUNT, 10, 0.03))
feature_params = dict(maxCorners=500, qualityLevel=0.3, minDistance=7, blockSize=7)
# test_areas = [] # 固定四个角
# test_areas.append((0, 0, 128, 128)) # lu
# test_areas.append((640 - 128, 0, 640, 128)) # ru
# test_areas.append((640 - 128, 480 - 128, 640, 480)) # rd
# test_areas.append((0, 480 - 128, 128, 480)) # ld
while True:
if cv2.waitKey(1) == 27: # Esc for exit
break
t = clock()
ret, img = cap.read()
img = cv2.flip(img, 1)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
h, w = gray.shape
rectangles = detect(gray, f_cascade)
# draw_rectangles(img, rectangles, color=(255, 225, 0)) # 人脸范围
# draw_rectangles(img, rectangles, scaling=1.3) # 扩大后的范围
if len(rectangles) == 1: # 限制一张人脸
for rectangle in rectangles:
rx0, ry0, rx1, ry1 = rectangle
# if rx1 - rx0 != 151 and ry1 - ry0 != 151: # 限定人脸识别框的大小
# continue
draw_rectangle(img, rectangle, color=(255, 225, 0)) # 人脸范围
# draw_rectangle(img, rectangle, scaling=1.4) # 扩大后的范围
rectangles_eye = detect(gray[ry0:ry1, rx0:rx1], e_cascade)
# if len(rectangles_eye) < 1: # 限制必须两只眼睛都识别出来
# continue
test_areas_face = []
size = 32
face_h = ry1 - ry0
face_w = rx1 - rx0
test_areas_face.append((0, 0, size, size)) # lu
test_areas_face.append((face_w - size, 0, face_w, size)) # ru
test_areas_face.append((face_w - size, face_h - size, face_w, face_h)) # rd
test_areas_face.append((0, face_h - size, size, face_h)) # ld
test_areas_face.append((face_w / 4, 0, face_w / 4 * 3, size)) # mu
test_areas_face.append((face_w / 4, face_h - size, face_w / 4 * 3, face_h)) # md
test_areas_face.append((0, face_h / 4, size, face_h / 4 * 3)) # ml
test_areas_face.append((face_w - size, face_h / 4, face_w, face_h / 4 * 3)) # mr
draw_rectangles(img[ry0:ry1, rx0:rx1], rectangles_eye, color=(255, 0, 225))
draw_rectangles(img[ry0:ry1, rx0:rx1], test_areas_face, color=(25, 23, 225))
eye_flow_lines = [] # 眼睛光流位移
eye_flow_angs = [] # 眼睛光流角度
for erx0, ery0, erx1, ery1 in rectangles_eye:
if len(tracks) > 0:
img0, img1 = prev_gray[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1], gray[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1]
p0 = np.float32([tr[-1] for tr in tracks]).reshape(-1, 1, 2)
p1, st, err = cv2.calcOpticalFlowPyrLK(img0, img1, p0, None, **lk_params)
p0r, st, err = cv2.calcOpticalFlowPyrLK(img1, img0, p1, None, **lk_params)
d = abs(p0 - p0r).reshape(-1, 2).max(-1)
good = d < 1
new_tracks = []
for tr, (x, y), good_flag in zip(tracks, p1.reshape(-1, 2), good):
if not good_flag:
continue
tr.append((x, y))
if len(tr) > track_len:
del tr[0]
new_tracks.append(tr)
cv2.circle(img[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1], (x, y), 2, (0, 255, 0), -1)
tracks = new_tracks
cv2.polylines(img[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1], [np.int32(tr) for tr in tracks], False, (0, 255, 0))
draw_str(img, (200, 20), 'track count: %d' % len(tracks))
if frame_idx % detect_interval == 0:
mask = np.zeros_like(gray[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1])
mask[:] = 255
for x, y in [np.int32(tr[-1]) for tr in tracks]:
cv2.circle(mask, (x, y), 2, 0, -1)
p = cv2.goodFeaturesToTrack(gray[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1], mask=mask, **feature_params)
if p is not None:
for x, y in np.float32(p).reshape(-1, 2):
tracks.append([(x, y)])
test_flow_lines = [] # 测试区域光流位移
test_flow_angs = [] # 测试区域光流角度
for erx0, ery0, erx1, ery1 in test_areas_face:
test_flow = opt_flow(prev_gray[ery0:ery1, erx0:erx1], gray[ery0:ery1, erx0:erx1]) # get opt flow
lines = draw_flow(img[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1], test_flow, step=4) # 显示光流点
test_flow_lines.append(lines)
# mag, ang = cv2.cartToPolar(eye_flow[..., 0], eye_flow[..., 1])
# test_flow_angs.append(ang)
eye_sorted = [] # 排序后的长度集合(眼睛)
for lines in eye_flow_lines:
mds = []
for (x1, y1), (x2, y2) in lines:
md = math.sqrt((x2 - x1) ** 2 + (y2 - y1) ** 2)
mds.append(md)
eye_sorted.append(sorted(mds, reverse=True))
test_sorted = [] # 排序后的长度集合(test)
for lines in test_flow_lines:
mds = []
for (x1, y1), (x2, y2) in lines:
md = math.sqrt((x2 - x1) ** 2 + (y2 - y1) ** 2)
mds.append(md)
test_sorted.append(sorted(mds, reverse=True))
# ========打印前十个=========================
if True:
for i, md2 in enumerate(eye_sorted):
count = 0
print '眼睛' + str(i + 1) + ':\t',
for md in md2:
count += 1
if count > 10:
break
print str(round(md, 2)) + ',',
print ''
print ''
for i, md2 in enumerate(test_sorted):
count = 0
print '测试' + str(i + 1) + ':\t',
for md in md2:
count += 1
if count > 10:
break
print str(round(md, 2)) + ',',
print ''
print ''
if False:
# ============= 打印前十个平均值 ============
for i, md2 in enumerate(eye_sorted):
count = 0
print '眼睛' + str(i + 1) + ':\t',
sum_avg = []
for md in md2:
count += 1
if count > 10:
break
sum_avg.append(md)
print round(1.0 * sum(sum_avg) / len(sum_avg), 2)
for i, md2 in enumerate(test_sorted):
count = 0
print '测试' + str(i + 1) + ':\t',
sum_avg = []
for md in md2:
count += 1
if count > 10:
break
sum_avg.append(md)
print round(1.0 * sum(sum_avg) / len(sum_avg), 2)
print ''
prev_gray = gray
dt = clock() - t
draw_str(img, (20, 20), 'time: %.1f ms' % (dt * 1000))
cv2.imshow("Face detect", img)
cap.release()
cv2.destroyAllWindows()
video_src = 0
args = dict(args)
face_fn = args.get('--face-cascade',
"data/haarcascades/haarcascade_frontalface_alt.xml")
hand_fn = args.get('--hand-cascade', "data/haarcascades/palm.xml")
face_cascade = cv2.CascadeClassifier(face_fn)
hand_cascade = cv2.CascadeClassifier(hand_fn)
ar_dictionary = cv2.aruco.getPredefinedDictionary(cv2.aruco.DICT_4X4_100)
cam = create_capture(video_src,
fallback='synth:bg=../data/lena.jpg:noise=0.05')
while True:
t = clock()
ret, img = cam.read()
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
cam_dt = clock() - t
t = clock()
rects = detect(gray, face_cascade, 80, 500)
vis = img.copy()
draw_rects(vis, rects, (0, 255, 0))
face_det_dt = clock() - t
t = clock()
x20 = 0
x21 = 0
timeout = 1
#=======================================================================
#starting while loop
#-------------------
while True:
global E1
ret, img = cam.read() #Readin image
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) #converting to gray scale
gray = cv2.equalizeHist(gray)
x30 = 0
E1 = 0
E2 = 0
t = clock()
rects = detect(gray, cascade) #activate function to detect face
vis = img.copy() #copying image
vistest = img.copy()
if not nested.empty():
for x1, y1, x2, y2 in rects:
roi = gray[y1:y2, x1:x2] #croping image
vis_roi = vis[y1:y2, x1:x2]
vistest_roi = vistest[y1:y2, x1:x2]
#cv2.imshow('face', vis_roi)
subrects = detect1(roi.copy(), nested) #detecting closed eye
subrects1 = detect2(roi.copy(),
nested1) #detecting closed and open eye
draw_rects(vistest_roi, subrects1, (255, 0, 0))
def Update(self):
print("camera start")
cv2.destroyAllWindows()
# initialize the camera and grab a reference to the raw camera capture
self._camera = PiCamera()
self._camera.resolution = (self.__cameraResolutionX,
self.__cameraResolutionY)
self._camera.framerate = 32
self._rawCapture = PiRGBArray(
self._camera
) #, size=(self._camera.resolution.width, self._camera.resolution.height))
# allow the camera to warmup
time.sleep(0.1)
cascade_fn = "/home/pi/opencv-3.1.0/data/haarcascades/haarcascade_frontalface_alt.xml"
nested_fn = "/home/pi/opencv-3.1.0/data/haarcascades/haarcascade_eye.xml"
#cascade_fn = args.get('--cascade', "../../data/haarcascades/haarcascade_frontalface_default.xml")
#nested_fn = args.get('--nested-cascade', "../../data/haarcascades/haarcascade_eye.xml")
self._cascade = cv2.CascadeClassifier(cascade_fn)
self._nested = cv2.CascadeClassifier(nested_fn)
for frame in self._camera.capture_continuous(self._rawCapture,
format="bgr",
use_video_port=True):
if (super().updating_ended == True):
return
# grab the raw NumPy array representing the image, then initialize the timestamp
# and occupied/unoccupied text
image = frame.array
# local modules
#from video import create_capture
from common import clock, draw_str
#self._camera.capture(self._rawCapture, format="bgr")
#image = self._rawCapture.array
cv2.imshow('image', image)
# clear the stream in preparation for the next frame
self._rawCapture.truncate(0)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects = self.detect(gray, self._cascade)
if (self._showImage == True):
vis = image.copy()
self.draw_rects(vis, rects, (0, 255, 0))
dt = 0
found_something = False
if not self._nested.empty():
posX = -1
posY = -1
bestWidth = -1
for x1, y1, x2, y2 in rects:
width = x2 - x1
if (width > bestWidth):
bestWidth = width
posX = (x1 +
(x2 - x1) / 2) / self._camera.resolution.width
posY = (y1 +
(y2 - y1) / 2) / self._camera.resolution.height
if (self._showImage == True):
roi = gray[y1:y2, x1:x2]
vis_roi = vis[y1:y2, x1:x2]
subrects = self.detect(roi.copy(), self._nested)
self.draw_rects(vis_roi, subrects, (255, 0, 0))
self.posXFace = posX
self.posYFace = posY
dt = clock() - t
if (posX != -1):
#print('camera time: %.1f ms' % (dt*1000))
found_something = True
if (self._showImage == True):
draw_str(vis, (20, 20), 'time: %.1f ms' % (dt * 1000))
cv2.imshow('facedetect', vis)
if (found_something == True):
time.sleep(self._delay_seconds)
else:
time.sleep(self._delay_seconds_when_idle)
import sys
try:
fn1 = sys.argv[1]
except:
fn1 = "test_images/mona_lisa_face.png"
print help_message
# Good features parameters
gf_params = dict(maxCorners=200,
qualityLevel=0.1,
minDistance=7,
blockSize=20,
useHarrisDetector=False,
k=0.04)
img = cv2.imread(fn1, cv2.CV_LOAD_IMAGE_COLOR)
grey = cv2.cvtColor(img, cv.CV_BGR2GRAY)
start = clock()
keypoints = cv2.goodFeaturesToTrack(grey, mask=None, **gf_params)
if keypoints is not None:
for x, y in np.float32(keypoints).reshape(-1, 2):
cv2.circle(img, (x, y), 3, (0, 255, 0, 0), cv.CV_FILLED, 8, 0)
print "Elapsed time:", 1000 * (clock() - start), "milliseconds"
cv2.imshow("Keypoints", img)
cv2.waitKey()
def evaluate_motempl_model(self, video_path1, video_path2, model_params):
camera = cv2.VideoCapture(video_path1)
camera2 = cv2.VideoCapture(video_path2)
assert camera.isOpened() and camera2.isOpened(
), 'Can not capture source!'
flow_data = []
img_data = []
input_data = []
# model_params['crop_size']=224
motion_history = np.zeros(
(model_params['crop_size'][0], model_params['crop_size'][1]),
np.float32)
action = 'Normal'
prob = 1.0
flow_data2 = []
img_data2 = []
input_data2 = []
action2 = 'Normal'
motion_history2 = np.zeros(
(model_params['crop_size'][0], model_params['crop_size'][1]),
np.float32)
prob2 = 1.0
while camera.isOpened() and camera2.isOpened():
try:
_, frame = camera.read()
_, frame2 = camera2.read()
temp_frame = cv2.resize(frame, (model_params['crop_size'][0],
model_params['crop_size'][1]),
interpolation=cv2.INTER_CUBIC)
temp_frame2 = cv2.resize(frame2,
(model_params['crop_size'][0],
model_params['crop_size'][1]),
interpolation=cv2.INTER_CUBIC)
img_data.append(temp_frame)
img_data2.append(temp_frame2)
# Calculate the motempl flow between two frames of camera1
if len(img_data) == 3:
timestamp = clock()
flow_img = self.calc_motempl(img_data[0], img_data[2],
motion_history, timestamp)
flow_img = flow_img * 1.0 / 127.5
cv2.imshow('mote1', flow_img)
flow_img = np.array(flow_img)
flow_data.append(flow_img)
img_data = []
# Calculate the motempl flow between two frames of camera2
if len(img_data2) == 3:
timestamp2 = clock()
flow_img2 = self.calc_motempl(img_data2[0], img_data2[2],
motion_history2, timestamp2)
flow_img2 = flow_img2 * 1.0 / 127.5
cv2.imshow('mote2', flow_img2)
flow_img2 = np.array(flow_img2)
flow_data2.append(flow_img2)
img_data2 = []
# camera1
if len(flow_data) == model_params['sequence_length']:
action, prob, _ = self.calc_output(flow_data, input_data)
flow_data = []
input_data = []
# camera2
if len(flow_data2) == model_params['sequence_length']:
action2, prob2, _ = self.calc_output(
flow_data2, input_data2)
flow_data2 = []
input_data2 = []
cv2.putText(frame, action, (20, 30), cv2.FONT_HERSHEY_SIMPLEX,
1, text_color[action], 3)
cv2.putText(frame, str(prob), (20, 90),
cv2.FONT_HERSHEY_SIMPLEX, 1, text_color[action], 3)
cv2.putText(frame2, action2, (20, 30),
cv2.FONT_HERSHEY_SIMPLEX, 1, text_color[action2],
3)
cv2.putText(frame2, str(prob2), (20, 90),
cv2.FONT_HERSHEY_SIMPLEX, 1, text_color[action2],
3)
cv2.imshow('camera1', frame)
cv2.imshow('camera2', frame2)
choice = cv2.waitKey(1)
choice = cv2.waitKey(1)
except Exception as e:
print(e)
camera = cv2.VideoCapture(video_path1)
camera2 = cv2.VideoCapture(video_path2)
def run(folder, file_name):
video_src = os.path.join(os.curdir, folder, file_name)
images_folder = os.path.join(folder, 'frames')
if not os.path.isdir(images_folder):
os.mkdir(images_folder)
if os.path.isfile(video_src):
print('it is a file: ', video_src)
else:
print('no file you n00b', video_src)
# args, video_src = getopt.getopt(sys.argv[1:], '', ['cascade=', 'nested-cascade='])
# try:
# video_src = video_src[0]
# except:
# video_src = 0
# args = dict(args)
# cascade_fn = args.get('--cascade', "haarcascade_frontalface_alt.xml")
cascade_fn = "haarcascade_frontalface_alt.xml"
# nested_fn = args.get('--nested-cascade', "haarcascade_eye.xml")
nested_fn = "haarcascade_eye.xml"
cascade = cv2.CascadeClassifier(cascade_fn)
nested = cv2.CascadeClassifier(nested_fn)
# cam = create_capture(video_src, fallback='synth:bg=../data/lena.jpg:noise=0.05')
cam = cv2.VideoCapture(video_src)
image_count = 0
frame_count = 0
while True:
if frame_count % 500 == 0:
print('faces found: ', image_count)
print(int(cam.get(cv2.CAP_PROP_POS_FRAMES)), ": frames read")
ret, img = cam.read()
if not ret:
break
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects = detect(gray, cascade)
vis = img.copy()
draw_rects(vis, rects, (0, 255, 0))
if not nested.empty():
for x1, y1, x2, y2 in rects:
roi = gray[y1:y2, x1:x2]
vis_roi = vis[y1:y2, x1:x2]
subrects = detect(roi.copy(), nested)
#If there is eyes on the picture
if len(subrects) > 0:
#for every 20 frame it should save a picture.
if int(cam.get(cv2.CAP_PROP_POS_FRAMES)) % 20 == 0:
save_file_to = os.path.join(
images_folder,
"{0}{1}.jpg".format(file_name[:-4], image_count))
print("Captured: ",
(images_folder, "{0}{1}.jpg".format(
file_name[:-4], image_count)))
print("At: ", cam.get(cv2.CAP_PROP_POS_MEC))
# print('saving file to: ', save_file_to)
cv2.imwrite(save_file_to, img)
image_count += 1
# draw_rects(vis_roi, subrects, (255, 0, 0))
dt = clock() - t
frame_count += 1
# draw_str(vis, (20, 20), 'time: %.1f ms' % (dt*1000))
#cv2.imshow('facedetect', vis)
if cv2.waitKey(5) == 27:
break
print("{0} images made out of {1} frames".format(image_count, frame_count))
cv2.destroyAllWindows()
def demo_video(video_file):
detector = ObjectDetectorYolo(model='tiny-yolo-voc')
mtracker = KalmanTracker(['person'], tracker='deep_sort')
cap = common.VideoStream(video_file, queueSize=4).start()
cv2.waitKey(500)
Outcount, Incount = 0, 0
total_t, counter = 0, 0
while not cap.stopped:
t = common.clock()
imgcv = cap.read()
if imgcv is not None:
counter += 1
detections = detector.run(imgcv)
mtracker.update(imgcv, detections)
cvboxes, ids = [], []
for tid, tracker in mtracker.trackers.iteritems():
if tracker.consecutive_invisible_count < 5:
state_current = get_pos(tracker.bbox)
try:
if state_current != tracker.regionside:
tracker.statechange += 1
print state_current, tracker.regionside, tracker.statechange
if state_current == 'Positive':
if tracker.statechange % 2:
Incount += 1
else:
Outcount -= 1
else:
if tracker.statechange % 2:
Outcount += 1
else:
Incount -= 1
tracker.regionside = state_current
except AttributeError:
tracker.regionside = state_current
tracker.statechange = 0
cvboxes.append(tracker.bbox)
ids.append(tid)
print Incount, Outcount
cv2.line(imgcv, (LINE['x1'], LINE['y1']), (LINE['x2'], LINE['y2']),
(0, 0, 255), 4)
common.drawLabel(imgcv,
"IN:%d OUT:%d" % (Incount, Outcount), (10, 10),
size=1,
color=(0, 0, 255))
common.showImage(draw_boxes(imgcv, cvboxes, ids))
key = cv2.waitKey(1) & 0xFF
if key == 27:
break
t1 = common.clock()
dt = t1 - t
t = t1
total_t += dt
print counter / total_t
def processFrame(self, frame, t0):
keypoints = None
lines = None
contours = None
cmode = self.cmode
values = self.values
if cmode != 'rgb':
if cmode in ['h', 's', 'v']:
#mode = (cv2.COLOR_BGR2HLS, cv2.COLOR_HLS2BGR)
mode = (cv2.COLOR_BGR2HSV, cv2.COLOR_HSV2BGR)
hsv = cv2.cvtColor(frame, mode[0])
if cmode == 'h':
if 0:
if mode[0] == cv2.COLOR_BGR2HSV:
hsv[:, :, 1] = 255 # s = 1
hsv[:, :, 2] = 128 # v = .5
else:
hsv[:, :, 1] = 128 # l = .5
hsv[:, :, 2] = 255 # s = 1
frame = cv2.cvtColor(hsv, mode[1])
else:
h, s, v = cv2.split(hsv)
# now find the interesting range of hues..
frame = cv2.inRange(h, values[0], values[1])
#frame = frame * v
elif cmode == 's':
# extract the s as grayscale
if mode[0] == cv2.COLOR_BGR2HSV:
h, frame, v = cv2.split(hsv)
else:
h, l, frame = cv2.split(hsv)
elif cmode == 'v':
if mode[0] == cv2.COLOR_BGR2HSV:
h, s, frame = cv2.split(hsv)
else:
h, frame, s = cv2.split(hsv)
elif cmode in ['r', 'g', 'b']:
if cmode == 'r':
b, g, frame = cv2.split(frame)
elif cmode == 'g':
b, frame, r = cv2.split(frame)
elif cmode == 'b':
frame, g, r = cv2.split(frame)
elif cmode == 'adaptiveThreshold':
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
if values[1] == 0:
thresh = cv2.THRESH_BINARY
else:
thresh = cv2.THRESH_BINARY_INV
frame = cv2.adaptiveThreshold(
gray,
200, # value to draw
cv2.ADAPTIVE_THRESH_MEAN_C,
thresh,
5,
values[0])
elif cmode == 'threshold':
frame = self.simpleThreshold(frame, values)
elif cmode == 'huerange*valrange':
frame = self.hvRange(frame)
elif cmode == 'canny':
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
v1 = values[0] * 200
v2 = v1 + values[1]
frame = cv2.Canny(frame, v1, v2, apertureSize=5)
elif cmode == 'simpleblob':
if not "blobdetector" in self.algostate or self.update:
bp = cv2.SimpleBlobDetector_Params()
if values[4] >= 0:
bp.filterByColor = True
bp.blobColor = values[4] # 0 or 255 (?)
else:
bp.filterByColor = False
bp.blobColor = 0
# Change thresholds
bp.minThreshold = values[0] # 50
bp.maxThreshold = values[1] # 150
bp.thresholdStep = values[2] # 5
# Filter by Area.
bp.filterByArea = True
bp.minArea = values[3] # 500
bp.maxArea = (640 * 480) / 5
# Filter by Circularity
bp.filterByCircularity = False
bp.minCircularity = 0.1
# Filter by Convexity
bp.filterByConvexity = False
bp.minConvexity = 0.87
# Filter by Inertia
bp.filterByInertia = False
bp.minInertiaRatio = 0.01
detector = cv2.SimpleBlobDetector(bp)
self.algostate["blobdetector"] = detector
else:
detector = self.algostate["blobdetector"]
if 0:
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
elif 0:
frame = self.hvRange(frame)
elif 0:
hvals = self.getCmodeValues('h')
hsv = cv2.cvtColor(frame, cv2.COLOR_BGR2HSV)
frame, s, v = cv2.split(hsv)
frame = cv2.inRange(frame, hvals[0], hvals[1])
else:
gvals = self.getCmodeValues('gamma')
gamma = 1 + 10 * gvals[0] / 100.0
self.putNotice('gamma: %f' % gamma)
for i in range(0, 256):
self.LUT[i] = 255 * ((i / 255.0)**gamma)
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
frame = cv2.LUT(gray, self.LUT)
keypoints = detector.detect(frame) # we'll draw them
keypoints = self.robotCnx.NewKeypoints(keypoints)
elif cmode == "houghlines":
cv = self.getCmodeValues('canny')
bwf = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
v1 = cv[0] * 200
v2 = v1 + cv[1]
bwf = cv2.Canny(bwf, v1, v2, apertureSize=5)
rho = max(values[0], 1)
theta = max(math.radians(values[1]), 0)
threshold = max(values[2], 1) # votes needed to accept a line
minlen = values[3]
maxgap = values[4]
lines = cv2.HoughLinesP(
bwf,
rho, # distance res of accum in pixels
theta,
threshold,
minLineLength=minlen,
maxLineGap=maxgap)
lines = self.robotCnx.NewLines(lines)
elif cmode == 'contours':
gray = self.simpleThreshold(frame)
mode = cv2.RETR_TREE #values[0]
method = cv2.CHAIN_APPROX_SIMPLE #values[1]
off = (values[2], values[3])
contours = cv2.findContours(gray, mode, method, offset=off)
# NB: contours may be tuple (contours, hierarchy)
elif cmode == 'ORB':
if not "ORB" in self.algostate or self.update:
sf = max(1, min(2, 1. + values[1] / 255.)) # [1->2]
self.algostate["ORB"] = cv2.ORB(nfeatures=values[0],
scaleFactor=sf,
nlevels=values[2],
patchSize=values[3],
edgeThreshold=values[3])
# could add:
# WTA_K: 2
# scoreType ORB_HARRIS_SCORE
# patchSize ~= edgeThreshold
orb = self.algostate["ORB"]
#keypoints,descrs = orb.detectAndCompute(frame, None)
#gray = self.simpleThreshold(frame)
gray = self.hvRange(frame)
keypoints = orb.detect(gray, None)
self.putNotice('ORB features: %d' % len(keypoints))
frame = gray
elif cmode == 'dance0':
t = common.clock() * 2 * math.pi / 5
x = math.sin(t) * values[0] / 27 * 320 + 320 #[0, 640]s/
kp = cv2.KeyPoint(x, 240, 10)
keypoints = self.robotCnx.NewKeypoints([kp])
elif cmode == 'dance1':
t = common.clock() * 2 * math.pi / 15
x = 22 * math.sin(t) + 320
#[0, 640]s/
y = 60 * (1 + math.sin(t)) + 240
# math.sin(t)*values[0]/27*320 + 240 #[0, 480]s/
kp = cv2.KeyPoint(x, y, 10)
keypoints = self.robotCnx.NewKeypoints([kp])
elif cmode == 'gamma':
# Since our inputs are normalized to [0, 1]
# we want a power > 1 to reduce darker shades.
# map values [0, 100] -> [1 - 10]
gamma = 1 + 10 * values[0] / 100.0
self.putNotice('gamma: %f' % gamma)
for i in range(0, 256):
self.LUT[i] = 255 * ((i / 255.0)**gamma)
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
frame = cv2.LUT(gray, self.LUT)
else:
print("unknown cmode: " + cmode)
return frame, t0, keypoints, lines, contours
def myapp(environ, start_response):
global ws
global cam
global tracker
global face_fn
global con_fn
global tri_fn
global tracker
global conns
global trigs
global cam
global shape3D
global ms
ws = environ['wsgi.websocket']
print('enter!')
ws.send("hoge!")
plot_flag = False
try:
while True:
t = clock()
ret, img = cam.read()
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
if tracker.update(gray):
draw_str(img, (20, 40), 'pos: %.1f, %.1f' % tracker.getPosition())
draw_str(img, (20, 60), 'scale: %.1f ' % tracker.getScale())
draw_str(img, (20, 80), 'orientation: %.1f, %.1f, %.1f' % tracker.getOrientation())
tracker.getScale()
tracker.getOrientation()
#img = tracker.draw(img, conns, trigs)
img = tracker.draw(img, conns, None)
shape3D = tracker.get3DShape().reshape((3, 66))
#print_cood(shape3D)
print(t)
# #left eye (check max - min on Y axis)
# if isClosed(shape3D[1][36:41].tolist(), 2) == True:
# print("left eye closed")
#
# #right eye (check max - min on Y axis)
# if isClosed(shape3D[1][42:47].tolist(), 2) == True:
# print("right eye closed")
# rip (check max - min on Y axis)
if isClosed(shape3D[1][60:65].tolist(), 0.7) == True:
print("rip closed")
if ws != None:
ws.send("close")
sleep(0.3)
else:
print("rip open")
if ws != None:
ws.send("open")
# comment out
#print shape3D.min(), shape3D.max()
#comment out because need less
#ms.set(x=shape3D[0, :] , y=shape3D[1, :], z=shape3D[2, :])
else:
tracker.setWindowSizes((11, 9, 7))
dt = clock() - t
draw_str(img, (20, 20), 'time: %.1f ms' % (dt*1000))
cv2.imshow('facedetect', img)
if 0xFF & cv2.waitKey(5) == 27:
break
except:
pass
cv2.destroyAllWindows()
def main(cam_id, c_type):
"""
由于眼部检测的不稳定性，这里采用在已经识别出的人脸范围内假定的眼睛区域作为眼睛区域的判断
采用 关键点检测 加 伪眼部光流检测
关键点静止但眼部光流明显时判定为活体
这个主要判断 眨眼 来做为活体检测的依据
更换为双摄像头
并判断
1、电子设备（普通摄像头中有人脸区域而红外摄像头中没有）
2、照片（眼睛区域关键点位移的方差与眼睛外部的方差 差值小于阈值时判定为照片，但有很大程度会把真人识别成照片
修改为判断连续 n 帧的方差差值，如果全部符合再判定为照片）
3、通过，活体（。。。）
:param cam_id: 摄像头ID
:param c_type: 类型（color:Visible Light 可见光, gray:infrared 红外）
:return:
"""
print(__doc__)
cap = cv2.VideoCapture(cam_id)
fourcc = cv2.VideoWriter_fourcc(*'XVID')
out = cv2.VideoWriter('D:/data.avi', fourcc, 30, (640, 480))
f_cascade = cv2.CascadeClassifier(
"C:/opencv/opencv/build/etc/haarcascades/haarcascade_frontalface_alt2.xml"
)
e_cascade = cv2.CascadeClassifier(
"C:\\opencv\\opencv\\build\etc\\haarcascades\\haarcascade_eye.xml")
ret, prev = cap.read()
prev_gray = cv2.cvtColor(prev, cv2.COLOR_BGR2GRAY)
lk_params = dict(winSize=(15, 15),
maxLevel=2,
criteria=(cv2.TERM_CRITERIA_EPS | cv2.TERM_CRITERIA_COUNT,
10, 0.03))
feature_params = dict(maxCorners=500,
qualityLevel=0.3,
minDistance=7,
blockSize=7)
tracks = []
frame_index = 0 # 总体的帧数， 用于判断电子设备
face_frame_index = 0 # 脸部区域的帧数，用于 关键点轨迹
detect_interval = 3
track_len = 10
msg_show_success = 0 # 包含光流和关键点判断通过的
msg_show_opt = 0 # 通过信息显示帧数
msg_show_key = 0 # 通过信息显示帧数
msg_show_success_f = 0 # 没有通过包含光流和关键点判断通过的
msg_show_opt_f = 0 # 没有通过通过信息显示帧数
msg_show_key_f = 0 # 没有通过通过信息显示帧数
has_face = False
sustain = 10 # 信息持续时间
ph_check_count = 30 # 判断为照片时 持续的帧数
photo_rec = [0] * ph_check_count # 记录每一帧的照片判定
hu_check_count = 30 # 判断为真人时 持续的帧数
human_rec = [0] * hu_check_count # 记录为真人
# 存储每一帧的光流
eye_flow_lines_t = []
while True:
if cv2.waitKey(1) == 27: # Esc for exit
break
t = clock()
ret, img = cap.read()
if c_type == 'gray':
img = img[24:456, 32:608]
img = cv2.resize(img, (640, 480), interpolation=cv2.INTER_LINEAR)
img = cv2.flip(img, 1)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
frame_index += 1
rectangles = detect(gray, f_cascade)
mask = np.zeros_like(gray) # 设置关键点遮罩
photo_rec[frame_index % ph_check_count] = 0
human_rec[frame_index % hu_check_count] = 0
if len(rectangles) == 1: # 限制一张人脸
share[c_type][frame_index % el_check_count] = 1 # 识别出人脸后将值设置为1
if not (has_face and True):
tracks = []
has_face = True
for rectangle in rectangles:
rx0, ry0, rx1, ry1 = rectangle
if rx1 - rx0 > 211:
draw_str(img, (20, 20), 'Close', front=(0, 0, 255))
elif rx1 - rx0 < 211:
draw_str(img, (20, 20), 'Away', front=(0, 0, 255))
else:
draw_str(img, (20, 20), 'OK. Hold.', front=(0, 255, 0))
draw_rectangle(img, rectangle, color=(0, 225, 0)) # 人脸范围
rectangles_eye = detect(gray[ry0:ry1, rx0:rx1],
e_cascade) # 获取眼睛范围
# draw_rectangles(img[ry0:ry1, rx0:rx1], rectangles_eye, color=(255, 0, 225))
# 眼部光流场功能
eye_flow_lines = []
# for erx0, ery0, erx1, ery1 in rectangles_eye:
# eye_flow = opt_flow(prev_gray[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1],
# gray[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1]) # get opt flow
# eye_flow_lines.append(draw_flow(img[ry0:ry1, rx0:rx1][ery0:ery1, erx0:erx1],
# eye_flow, step=4)) # 显示光流点
# 假眼部位置，假设脸纵向上部1/4位置到2/4位置及 横向左部1/6到5/6位置为眼部，以抵消眼部识别不能每次都有效地问题
face_h = ry1 - ry0
face_w = rx1 - rx0
face_hs = face_h / 4
face_he = face_h / 2
face_ws = face_w / 6
face_we = face_w / 6 * 5
eye_flow = opt_flow(
prev_gray[ry0:ry1, rx0:rx1][face_hs:face_he,
face_ws:face_we],
gray[ry0:ry1, rx0:rx1][face_hs:face_he, face_ws:face_we])
eye_flow_lines.append(
draw_flow(img[ry0:ry1, rx0:rx1][face_hs:face_he,
face_ws:face_we],
eye_flow,
step=4))
eye_sorted = [] # 排序后的长度集合(眼睛)
eye_sorted2 = []
for lines in eye_flow_lines:
mds = []
for (x1, y1), (x2, y2) in lines:
md = math.sqrt((x2 - x1)**2 + (y2 - y1)**2)
mds.append(md)
eye_sorted2.append(md)
eye_sorted.append(sorted(mds, reverse=True))
eye_flow_lines_t.append(eye_sorted2) # 存储每一帧的光流位移信息
# 绘制关键点轨迹
# 会删除位移较大的关键点
# 不影响其他的鉴别
if len(tracks) > 0:
img0, img1 = prev_gray, gray
p0 = np.float32([tr[-1]
for tr in tracks]).reshape(-1, 1, 2)
p1, st, err = cv2.calcOpticalFlowPyrLK(
img0, img1, p0, None, **lk_params)
p0r, st, err = cv2.calcOpticalFlowPyrLK(
img1, img0, p1, None, **lk_params)
d = abs(p0 - p0r).reshape(-1, 2).max(-1)
good = d < 0.5
new_tracks = []
for tr, (x, y), good_flag in zip(tracks, p1.reshape(-1, 2),
good):
if not good_flag:
continue
if not (rx0 < x < rx1 and ry0 < y < ry1):
continue
tr.append((x, y))
if len(tr) > track_len:
del tr[0]
new_tracks.append(tr)
cv2.circle(img, (x, y), 2, (0, 255, 0), -1)
tracks = new_tracks
cv2.polylines(img, [np.int32(tr) for tr in tracks], False,
(0, 255, 0))
# draw_str(img, (20, 20), 'track count: %d' % len(tracks))
# 限定人脸为兴趣区域
cv2.fillPoly(
mask,
np.array([[[rx0, ry0], [rx1, ry0], [rx1, ry1],
[rx0, ry1]]]), (255, 255, 255))
for x, y in [np.int32(tr[-1]) for tr in tracks]:
cv2.circle(mask, (x, y), 5, 0, -1) # 排除上一次的关键点
if face_frame_index % detect_interval == 0:
# print('**************** start ***************')
l_sorted = []
l_sorted_eye = [] # 眼睛区域的关键点
l_sorted_out = [] # 眼睛外部的关键点
l_tmp = []
l_tmp_eye = []
l_tmp_out = []
for tr in tracks:
(x0, y0) = tr[0]
(x1, y1) = tr[-1]
if rx0 + face_ws < x1 < rx0 + face_we and ry0 + face_hs < y1 < ry1 + face_he:
l_tmp_eye.append(
round(math.sqrt((x1 - x0)**2 + (y1 - y0)**2),
2))
else:
l_tmp_out.append(
round(math.sqrt((x1 - x0)**2 + (y1 - y0)**2),
2))
l = round(math.sqrt((x1 - x0)**2 + (y1 - y0)**2), 2)
l_tmp.append(l)
# if l > 0:
# print(round(math.atan(abs((y1 - y0) / (x1 - x0))) / math.pi * 180, 2), end=':')
# print(l, end='\t')
# print('\n+++++++++++++++')
l_sorted = sorted(l_tmp, reverse=True)
l_sorted_eye = sorted(l_tmp_eye, reverse=True)
l_sorted_out = sorted(l_tmp_out, reverse=True)
if len(l_sorted_eye) > 0:
print(l_sorted_eye[0])
if len(l_sorted_out) > 0:
print(l_sorted_out[0])
print("--------------")
if len(l_sorted_out) > 3 and len(l_sorted_eye) > 3 \
and l_sorted_out[0] < 1 and l_sorted_eye[0] > 1 \
and l_sorted_eye[0] - l_sorted_out[0] > 1:
# print('xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx')
msg_show_key = sustain
msg_show_success = sustain
# human_rec[frame_index % hu_check_count] = 1
elif (len(l_sorted_out) > 3 and len(l_sorted_eye) > 3 and abs(np.var(l_sorted_eye[:3]) - np.var(l_sorted_out[:3])) < 0.0005)\
or (len(l_sorted_eye) > 1 and len(l_sorted_out) > 1 and l_sorted_out[0] < 0.1 and l_sorted_eye[0] < 0.1):
print(
np.var(l_sorted_eye[:3]) -
np.var(l_sorted_out[:3]))
print("yesyesyesyesyesyes")
# 判定照片
# msg_show_key_f = sustain
# msg_show_success_f = sustain
photo_rec[frame_index % ph_check_count] = 1
# ========打印前十个=========================
# if True:
# for i, md2 in enumerate(eye_sorted):
# count = 0
# print('眼睛', str(i + 1), end=':\t')
# for md in md2:
# count += 1
# if count > 150:
# break
# print(round(md, 2), end=',')
# print()
# print('###################')
# 活体检测
np_eye = np.array(sorted(eye_sorted2, reverse=True)[:30])
np_eye = np_eye[np_eye > 0]
np_l = np.array(l_sorted[:10])
# print(np_eye.size, '+++++', np_l.size)
if np_eye.size != 0 and np_l.size != 0:
flow_pre = np_eye[np_eye > 2].size * 1.0 / np_eye.size
ln_pre = np_l[np_l > 2].size * 1.0 / np_l.size
# print(flow_pre, '---', ln_pre)
if 0.8 > flow_pre > 0.05 and ln_pre < 0.2:
msg_show_opt = sustain
msg_show_success = sustain
# print('yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy')
# elif flow_pre < 0.05 and ln_pre < 0.2:
# msg_show_opt_f = sustain
# msg_show_success_f = sustain
# print('**************** end ***************')
# 判断关键点
p = cv2.goodFeaturesToTrack(gray,
mask=mask,
**feature_params)
if p is not None:
for x, y in np.float32(p).reshape(-1, 2):
tracks.append([(x, y)])
face_frame_index += 1
else:
has_face = False
share[c_type][frame_index %
el_check_count] = 0 # 没识别出人脸（或识别出多个）后将值设置为0
prev_gray = gray
dt = clock() - t
# draw_str(img, (20, 20), 'time: %.1f ms' % (dt * 1000))
# if msg_show_key > 0:
# draw_str(img, (450, 20), 'YES by KEY', front=(0, 0, 255))
# msg_show_key -= 1
# if msg_show_opt > 0:
# draw_str(img, (300, 20), 'YES by OPT', front=(0, 0, 255))
# if c_type == 'color'> 0:
# print(sum(photo_rec))
if sum(share['color']) > el_check_count * 0.99 and sum(
share['gray']) == 0: # color中80%的帧里面识别出人脸，gray中大于80%的帧中没有识别出人脸
draw_str(img, (400, 30), 'Electronic', front=(0, 0, 255), size=2)
msg_show_success = 0
msg_show_success_f = 0
elif sum(photo_rec) > ph_check_count * 0.1:
draw_str(img, (400, 30), 'Photo', front=(0, 0, 255), size=2)
msg_show_success = 0
elif sum(share['color']) > el_check_count * 0.99 \
and msg_show_success > 0: # and sum(human_rec) > 0:
draw_str(img, (400, 30), 'Pass', front=(0, 255, 0), size=2)
if msg_show_success > 0:
msg_show_success -= 1
# msg_show_success_f -= 1
# if c_type == 'color':
cv2.imshow(c_type + str(cam_id), img)
out.write(img)
cap.release()
out.release()
cv2.destroyAllWindows()
low_image = shm.zeros(img.shape[0:2], dtype=img.dtype)
# controller_thread = threading.Thread(target=controller, args=(controller_state, stopping, controller_cv, the_lock))
# controller_thread = mp.Process(target=controller, args=(controller_state, stopping, controller_cv, the_lock))
# controller_thread.start()
targeter = Targeter(height, width, add_controller_command)
primed = True
firing = False
recognizing = False
locked_counter = 0
with recog.RecognizerManager() as recognizer:
while True:
t = clock()
ret, img = cam.read()
with the_lock:
low_image[:] = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
low_image[:] = cv2.equalizeHist(low_image)
if high_proc is None:
high_proc = mp.Process(target=high_level_tracker,
args=(
low_image,
stopping,
the_lock,
possible_targets,
cascade,
))
yServo = xServo servoPoint = Point(int(servo_min+((servo_max - servo_min)/2)), int(servo_min+((servo_max - servo_min)/2))) lastFacePoint=0 lastPIDUpdate=0 servoPrevPoint = Point(int(servo_min+((servo_max - servo_min)/2)), int(servo_min+((servo_max - servo_min)/2))) servoSpeed = Point(10,10) xServoID = int(1) yServoID = int(0) curRect=-1 rectStart=clock() dt = 0 t=0 def exit_signal(signum, frame): global isRunning isRunning=False signal.signal(signal.SIGINT, exit_signal) signal.signal(signal.SIGTERM, exit_signal) def rectsToPoint(rects, point): if len(rects) >= 1: curRect = 0 point.rectToPoint(rects[0]) """if len(rects) == 1:
subfolder = os.path.join(path, 'out_img/', name.split('.')[0])
if not os.path.exists(subfolder):
print subfolder
print '--------path not exist---------'
os.mkdir(subfolder)
i = -1
while True:
i = i + 1
#rint i
ret, img = cam.read()
#print ret
if not ret: break
if i % 10 == 0:
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects = detect(gray, cascade)
vis = img.copy()
if not nested.empty():
for x1, y1, x2, y2 in rects:
roi = gray[y1:y2, x1:x2]
vis_roi = vis[y1:y2, x1:x2]
res = cv2.resize(vis_roi, (48, 48),
interpolation=cv2.INTER_CUBIC)
im = Image.fromarray(res).convert('L')
def go(self):
# args is argparser.Namespace: an object whose members are
# accessed as: self.args.cannedimages
self.robotCnx = robotCnx.RobotCnx(self.args.fakerobot)
self.vsrc = None
if self.args.cannedimages:
self.fnpat = '../../pictures/RealFullField/%d.jpg'
else:
self.fnpat = None
self.index = 0
self.lastFn = None
self.mainImg = None
self.algostate = {} # stores persistant objects depending on cmode
self.threadn = cv2.getNumberOfCPUs()
self.pool = ThreadPool(processes=self.threadn)
self.pending = deque()
self.threadedMode = True
self.update = False
self.latency = common.StatValue()
self.frameT = common.StatValue()
self.lastFrameTime = common.clock()
self.lastStashTime = 0
self.stashFilename = "/var/tmp/imgServer.home/currentImage.jpg"
self.stashParams = [int(cv2.IMWRITE_JPEG_QUALITY), 50]
self.zeros = (0, 0, 0, 0, 0, 0)
self.LUT = np.array(range(0, 256)).astype('uint8')
self.indent = ' ' * 50
self.lastStatus = ""
# cmodelist maps a number [0, len] to an cmode/algorithm
self.cmodelist = [
'rgb',
'adaptiveThreshold',
'threshold',
'huerange*valrange',
'canny',
'simpleblob',
'houghlines',
'contours',
'ORB',
'dance1',
'gamma',
'b',
'g',
'h',
'r',
's',
'v',
]
# cmodeValueCache stores the edited values associated with each cmode.
# It starts empty, then builds up state as cmode changes.
self.cmodeValueCache = {
'adaptiveThreshold': [-4, 0, 0, 0, 0, 0],
'threshold': [75, 0, 0, 0, 0, 0],
'huerange*valrange': [55, 100, 255, 255, 0, 0], # works for g LED
'canny': [10, 200, 0, 0, 0, 0],
'simpleblob': [75, 150, 20, 40**2, 0, 0],
# minThresh
# maxThresh
# thresStep
# minSize (maxsize is 'large')
'houghlines': [2, 5, 10, 30, 2, 0],
# rho is distance resolution in pixels
# theta is angle in degress (larger -> more lines)
# threshold is measured in 'votes', higher -> fewer
# minLineLength
# maxLineGap
'contours': [1, 0, 0, -1, 1, 0],
# mode: [0-3]
# method: [CHAIN_APPROX_NONE,_SIMPLE,_TC89_L1, KOS]
# offset shift for contour
# unused
# which contour to draw, -1 is all
# which level of the contours to draw
'ORB': [40, 10, 8, 31, 0, 0],
# nfeatures
# scaleFactor [1 -> 2 (0->255)]
# nlevels
# patchSizea == edgeThreshold
'dance1': [20, 20, 0, 0, 0, 0],
#X(degrees)
#Y(degrees)
#Not rotating correctly (not interpreting correctly)
'gamma': [34, 0, 0, 0, 0, 0],
#alpha
#beta
'r': self.zeros,
'g': self.zeros,
'b': self.zeros,
'h': [0, 61, 0, 0, 0, 0],
's': self.zeros,
'v': self.zeros,
'rgb': self.zeros
}
#keyToCmode maps a key to a cmod
self.keyToCmode = {
49: self.cmodelist[1], # adaptiveThreshold on '1' key
50: self.cmodelist[2],
51: self.cmodelist[3],
52: self.cmodelist[4],
53: self.cmodelist[5],
54: self.cmodelist[6], # houghlines
55: self.cmodelist[7], # contours
56: self.cmodelist[8], # ORB
57: self.cmodelist[9], # dance1
48: self.cmodelist[10], # 0 key: gamma
98: 'b',
99: 'rgb',
103: 'g',
104: 'h',
114: 'r',
115: 's',
118: 'v',
}
self.keyToFrameChange = {
81: 'b',
83: 'n',
}
self.algoValueChange = {
127: ("reset", self.zeros), # keypad clear
190: ("v1Down", (-1, 0, 0, 0, 0, 0)), # f1
191: ("v1Up ", (1, 0, 0, 0, 0, 0)), # f2
84: ("v1Down", (-1, 0, 0, 0, 0, 0)), # downarrow
82: ("v1Up ", (1, 0, 0, 0, 0, 0)), # uparrow
191: ("v1Up ", (1, 0, 0, 0, 0, 0)), # downarrow
192: ("v2Down", (0, -1, 0, 0, 0, 0)), # f3
193: ("v2Up ", (0, 1, 0, 0, 0, 0)), # f4
194: ("v3Down", (0, 0, -1, 0, 0, 0)), # f5
195: ("v3Up ", (0, 0, 1, 0, 0, 0)), # f6
196: ("v4Down", (0, 0, 0, -1, 0, 0)), # f7
197: ("v4Up ", (0, 0, 0, 1, 0, 0)), # f8
198: ("v5Down", (0, 0, 0, 0, -1, 0)), # f9
199: ("v5Up ", (0, 0, 0, 0, 1, 0)), # f10
200: ("v6Down", (0, 0, 0, 0, 0, -1)), # f11
201: ("v6Up ", (0, 0, 0, 0, 0, 1)), # f12
}
self.cmode = self.getCmode(self.args.algorithm)
self.values = self.getCmodeValues(self.cmode)
def evaluate_motempl_model(self, video_path1, video_path2, model_params):
camera = cv2.VideoCapture(video_path1)
camera2 = cv2.VideoCapture(video_path2)
assert camera.isOpened() and camera2.isOpened(), 'Can not capture source!'
flow_data = []
img_data = []
input_data = []
motion_history = np.zeros((224, 224), np.float32)
action = 'Normal'
prob = 0.0
flow_data2 = []
img_data2 = []
input_data2 = []
action2 = 'Normal'
motion_history2 = np.zeros((224, 224), np.float32)
prob2 = 0.0
while camera.isOpened() and camera2.isOpened():
try:
_, frame = camera.read()
_, frame2 = camera2.read()
temp_frame = cv2.resize(frame, (224, 224),
interpolation=cv2.INTER_CUBIC)
temp_frame2 = cv2.resize(frame2, (224, 224),
interpolation=cv2.INTER_CUBIC)
img_data.append(temp_frame)
img_data2.append(temp_frame2)
# Calculate the optical flow between two frames of camera1
if len(img_data) == 2:
# flow_img = self.calc_motempl(img_data[0], img_data[1], motion_history)
frame_diff = cv2.absdiff(img_data[0], img_data[1])
gray_diff = cv2.cvtColor(frame_diff, cv2.COLOR_BGR2GRAY)
_, motion_mask = cv2.threshold(gray_diff, 32, 1, cv2.THRESH_BINARY)
timestamp = clock()
cv2.motempl.updateMotionHistory(motion_mask, motion_history, timestamp, MHI_DURATION)
flow_img = np.uint8(np.clip((motion_history - (timestamp - MHI_DURATION)) / MHI_DURATION, 0, 1) * 255)
flow_img = cv2.cvtColor(flow_img, cv2.COLOR_GRAY2BGR)
cv2.imshow('mote1', flow_img)
flow_img = flow_img * 1.0 / 127.5
flow_img = np.array(flow_img)
flow_data.append(flow_img)
img_data = []
# Calculate the optical flow between two frames of camera2
if len(img_data2) == 2:
# flow_img2 = self.calc_motempl(img_data2[0], img_data2[1], motion_history2)
# flow_img2 = flow_img2 * 1.0 / 127.5
frame_diff2 = cv2.absdiff(img_data2[0], img_data2[1])
gray_diff2 = cv2.cvtColor(frame_diff2, cv2.COLOR_BGR2GRAY)
_, motion_mask2 = cv2.threshold(gray_diff2, 32, 1, cv2.THRESH_BINARY)
timestamp2 = clock()
cv2.motempl.updateMotionHistory(motion_mask2, motion_history2, timestamp2, MHI_DURATION)
flow_img2 = np.uint8(np.clip((motion_history2 - (timestamp2 - MHI_DURATION)) / MHI_DURATION, 0, 1) * 255)
flow_img2 = cv2.cvtColor(flow_img2, cv2.COLOR_GRAY2BGR)
flow_img2 = flow_img2 * 1.0 / 127.5
cv2.imshow('mote2', flow_img2)
flow_img2 = np.array(flow_img2)
flow_data2.append(flow_img2)
img_data2 = []
# camera1
if len(flow_data) == model_params['sequence_length']:
action, prob, _ = self.calc_output(flow_data, input_data)
flow_data = []
input_data = []
# camera2
if len(flow_data2) == model_params['sequence_length']:
action2, prob2, _ = self.calc_output(flow_data2, input_data2)
flow_data2 = []
input_data2 = []
cv2.putText(frame, action, (20, 30), cv2.FONT_HERSHEY_SIMPLEX,
1, text_color[action], 3)
cv2.putText(frame, str(prob), (20, 90), cv2.FONT_HERSHEY_SIMPLEX,
1, text_color[action], 3)
cv2.putText(frame2, action2, (20, 30), cv2.FONT_HERSHEY_SIMPLEX,
1, text_color[action2], 3)
cv2.putText(frame2, str(prob2), (20, 90), cv2.FONT_HERSHEY_SIMPLEX,
1, text_color[action2], 3)
cv2.imshow('camera1', frame)
cv2.imshow('camera2', frame2)
choice = cv2.waitKey(10)
choice = cv2.waitKey(10)
except Exception as e:
print(e)
camera = cv2.VideoCapture(video_path1)
camera2 = cv2.VideoCapture(video_path2)
libunsharp.pool_init()
liblaplacian_naive.pool_init()
liblaplacian.pool_init()
libbilateral_naive.pool_init()
libbilateral.pool_init()
# Thickness setting has been moved from openCV to openCV 2.
# So, just use the raw value for CV_FILLED = -1
CV_THICKNESS_FILLED = -1
while (cap.isOpened()):
ret, frame = cap.read()
frameStart = clock()
rows = frame.shape[0]
cols = frame.shape[1]
if harris_mode:
if cv_mode:
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
gray = np.float32(gray) / 4.0
res = cv2.cornerHarris(gray, 3, 3, 0.04)
else:
res = np.empty((rows, cols), np.float32)
if naive_mode:
harris_naive(ctypes.c_int(cols - 2), ctypes.c_int(rows - 2),
ctypes.c_void_p(frame.ctypes.data),
ctypes.c_void_p(res.ctypes.data))
else:
harris(ctypes.c_int(cols - 2), ctypes.c_int(rows - 2),
xServo = servo_min + ((servo_max - servo_min) / 2)
yServo = xServo
servoPoint = Point(servo_min + ((servo_max - servo_min) / 2),
servo_min + ((servo_max - servo_min) / 2))
servoPrevPoint = Point(servo_min + ((servo_max - servo_min) / 2),
servo_min + ((servo_max - servo_min) / 2))
servoSpeed = Point(3, 3)
xServoID = 0
yServoID = 1
curRect = -1
rectStart = clock()
dt = 0
t = 0
def rectsToPoint(rects, point):
if len(rects) >= 1:
curRect = 0
point.rectToPoint(rects[0])
"""if len(rects) == 1:
curRect = 0
point.rectToPoint(rects[0])
elif len(rects) > curRect:
rectDt = (rectStart - clock()) * 1000
if rectDt > 10
curRect=int(round(random.random*(len(rects)-1)))
cascade = cv.CascadeClassifier(cv.samples.findFile(cascade_fn))
nested = cv.CascadeClassifier(cv.samples.findFile(nested_fn))
cam = create_capture(video_src,
fallback='synth:bg={}:noise=0.05'.format(
cv.samples.findFile('samples/data/lena.jpg')))
while True:
ret, img = cam.read() ##read the frame with the camera
gray = cv.cvtColor(
img, cv.COLOR_BGR2GRAY) ## turn the frame from RPG to GRAY
gray = cv.equalizeHist(gray) ##optimize the frame
vis = img.copy() ## make a copy of the frame
t = clock()
find_time = 0.0 ## the time we use to find a face in each frame, if this arg = 0, it means that we cant find a face in frame
# if we haven't a coordinate of the redrect.or we cant find a face in 2000ms,
# we scan the whole frame first
if Redrect.all() == 0 or total_time > 2000:
rects = detect(gray,
cascade) ## detect the faces from the whole frame
if len(rects):
Redrect = rects[0]
# if we find the faces, we update the data of the Redrect,
# and the area of the Redrect is 100 pixel bigger than the rect of the face
Redrect = np.array([
Redrect[0] - 100, Redrect[1] - 100, Redrect[2] + 100,
Redrect[3] + 100
])
cv.samples.findFile(cascade_fn)) # get the classification for the face
#nested = cv.CascadeClassifier(cv.samples.findFile(nested_fn)) # get the classification for eyes
cam = create_capture(video_src,
fallback='synth:bg={}:noise=0.05'.format(
cv.samples.findFile('samples/data/lena.jpg')))
#read the video and ready for the detecting
# Make a loop so that make up-to-date dections automatically
while rects == []:
ret, img = cam.read() # read the image
gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY) #make the image to grey
gray = cv.equalizeHist(gray) # make the histogram balance
t = clock() # mark down the starting time
rects = detect(gray, cascade) # capture the face area
dt = clock() - t # the time for the detecting
#make a copy of image and then can show the new one
vis = img.copy()
draw_rects(vis, rects, (0, 255, 0)) # show the image with frame
draw_str(vis, (20, 20),
'time: %.1f ms' % (dt * 1000)) # show the timing
cv.imshow('facedetect-camshift', vis) # show the plot
if cv.waitKey(
5
) == 27: # delay for the vesion things and control the breakpoint here
break
#stream = cv2.VideoWriter("/home/chekkaa/git/rob421-applied-robotics/stream.avi", cv.CV_FOURCC(*'MJPG'), 60.0, (640, 480), True)
shapes = [] # List of tracked shapes
cv2.namedWindow('binary')
cv2.namedWindow('contours')
cv2.namedWindow('raw')
cv2.createTrackbar('minarea', 'raw', 100, 10000, nothing)
cv2.createTrackbar('maxarea', 'raw', 800, 10000, nothing)
cv2.createTrackbar('lowthresh', 'raw', 200, 255, nothing)
cv2.createTrackbar('highthresh', 'raw', 255, 255, nothing)
while True:
ret, img = cam.read()
t = clock() # Start timing how long it took to process this frame
grayimg = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
grayimg = cv2.equalizeHist(grayimg)
lowthresh = cv2.getTrackbarPos('lowthresh', 'raw')
highthresh = cv2.getTrackbarPos('highthresh', 'raw')
ret, binimg = cv2.threshold(grayimg, lowthresh, highthresh, cv2.THRESH_BINARY)
#binimg = cv2.adaptiveThreshold(grayimg, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, 13, 5)
#binimg = cv2.Canny(grayimg, 250, 255)
# do erosion and dilation?
visbinimg = binimg.copy()
contours, hierarchy = cv2.findContours(binimg, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
cup_contours = []
for contour in contours:
contour_area = cv2.contourArea(contour)
minarea = cv2.getTrackbarPos('minarea', 'raw')
try:
video_src = video_src[0]
except:
video_src = 1
args = dict(args)
cascade_fn = args.get('--cascade', ".haarcascade_frontalface_alt2.xml")
cascade = cv2.CascadeClassifier(cascade_fn)
cam = create_capture(video_src)
while True:
ret, img = cam.read()
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.equalizeHist(gray)
t = clock()
rects = detect(gray, cascade)
vis = img.copy()
draw_rects(vis, rects, (0, 255, 0))
for x1, y1, x2, y2 in rects:
roi = gray[y1:y2, x1:x2]
vis_roi = vis[y1:y2, x1:x2]
dt = clock() - t
draw_str(vis, (20, 20), 'Processing time: %.1f ms' % (dt * 1000))
cv2.imshow('Face detection', vis)
cv2.imshow('Gray detection', gray)
if 0xFF & cv2.waitKey(5) == 27:
break
cv2.destroyAllWindows()
) # get trained face detected data
nested_fn = args.get('--nested-cascade',
"data/haarcascades/haarcascade_eye.xml")
cascade = cv.CascadeClassifier(
cv.samples.findFile(cascade_fn)) # Loads the classifier from a file.
nested = cv.CascadeClassifier(cv.samples.findFile(nested_fn))
cam = create_capture(
video_src, # create_capture is a convenience function for capture creation,
# falling back to procedural video in case of error.
fallback='synth:bg={}:noise=0.05'.format(
cv.samples.findFile('samples/data/lena.jpg')))
while True:
t = clock() # start to count the time
time = 0.0 # initialize the time counting
ret, img = cam.read() # capture a frame and store it
gray = cv.cvtColor(img, cv.COLOR_BGR2GRAY
) # convert the color of the image from bgr to gray
gray = cv.equalizeHist(
gray) # Equalizes the histogram of a grayscale image.
vis = img.copy() # make a copy of the image
if detected.all(
) == 0 or total_time > 3000: # if we haven't detected a face or can't find face in
# the current sub rectangle more than 3000ms,start to detect in the full image
rects = detect(gray, cascade)
if len(
rects
): # if find a face, create the sub rectangle by add 50 pixel for each x and y
# and initialize the total time
