def init(self): #Load the cascade classifier data self.cascade=cv.cvLoadHaarClassifierCascade(self.file_path(self.cascade_file),cv.cvSize(self.haar_face_size[0],self.haar_face_size[1])) #Allocate/init storage self.storage=cv.cvCreateMemStorage(0) cv.cvClearMemStorage(self.storage)
def detect_faces(self, img_grey): """ Detect faces within an image, then draw around them. The default parameters (scale_factor=1.1, min_neighbors=3, flags=0) are tuned for accurate yet slow object detection. For a faster operation on real video images the settings are: scale_factor=1.2, min_neighbors=2, flags=CV_HAAR_DO_CANNY_PRUNING, min_size=<minimum possible face size """ min_size = cv.cvSize(20,20) self.image_scale = 1.3 haar_scale = 1.2 min_neighbors = 2 haar_flags = 0 # Create a small image for better performance small_size = cv.cvSize(cv.cvRound(img_grey.width/self.image_scale),cv.cvRound(img_grey.height/self.image_scale)) small_img = cv.cvCreateImage(small_size, 8, 1) cv.cvResize(img_grey, small_img, cv.CV_INTER_LINEAR) cv.cvEqualizeHist(small_img, small_img) cv.cvClearMemStorage(self.faces_storage) if(self.cascade): t = cv.cvGetTickCount(); faces = cv.cvHaarDetectObjects(small_img, self.cascade, self.faces_storage, haar_scale, min_neighbors, haar_flags, min_size) t = cv.cvGetTickCount() - t cv.cvReleaseImage(small_img) #print "detection time = %gms" % (t/(cvGetTickFrequency()*1000.)); return faces
def face_detect(file, closeafter=True):
"""Converts an image to grayscale and prints the locations of any faces found"""
if hasattr(file, 'read'):
_, filename = tempfile.mkstemp()
tmphandle = open(filename, 'w')
shutil.copyfileobj(file, tmphandle)
tmphandle.close()
if closeafter:
file.close()
deleteafter = True
else:
filename = file
deleteafter = False
image = cvLoadImage(filename)
grayscale = cvCreateImage(cvSize(image.width, image.height), 8, 1)
cvCvtColor(image, grayscale, CV_BGR2GRAY)
storage = cvCreateMemStorage(0)
cvClearMemStorage(storage)
cvEqualizeHist(grayscale, grayscale)
cascade = cvLoadHaarClassifierCascade(
'/usr/share/opencv/haarcascades/haarcascade_frontalface_default.xml',
cvSize(1,1))
faces = cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2,
CV_HAAR_DO_CANNY_PRUNING, cvSize(50,50))
if deleteafter:
os.unlink(filename)
return (image.width, image.height), faces
def detectObject(image):
grayscale = cv.cvCreateImage(size, 8, 1)
cv.cvFlip(image, None, 1)
cv.cvCvtColor(image, grayscale, cv.CV_BGR2GRAY)
storage = cv.cvCreateMemStorage(0)
cv.cvClearMemStorage(storage)
cv.cvEqualizeHist(grayscale, grayscale)
cascade = cv.cvLoadHaarClassifierCascade(haar_file, cv.cvSize(1,1))
objects = cv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2,
cv.CV_HAAR_DO_CANNY_PRUNING,
cv.cvSize(100,100))
# Draw dots where hands are
if objects:
for i in objects:
#cv.cvRectangle(image, cv.cvPoint( int(i.x), int(i.y)),
# cv.cvPoint(int(i.x+i.width), int(i.y+i.height)),
# cv.CV_RGB(0,255,0), 3, 8, 0)
center = cv.cvPoint(int(i.x+i.width/2), int(i.y+i.height/2))
cv.cvCircle(image, center, 10, cv.CV_RGB(0,0,0), 5,8, 0)
# Left side check
if center.x > box_forward_left[0].x and center.x < box_backwards_left[1].x and center.y > box_forward_left[0].y and center.y < box_backwards_left[1].y:
set_speed('left', center)
# Right side check
if center.x > box_forward_right[0].x and center.x < box_backwards_right[1].x and center.y > box_forward_right[0].y and center.y < box_backwards_right[1].y:
set_speed('right', center)
def detectObject(image):
grayscale = cv.cvCreateImage(size, 8, 1)
cv.cvFlip(image, None, 1)
cv.cvCvtColor(image, grayscale, cv.CV_BGR2GRAY)
storage = cv.cvCreateMemStorage(0)
cv.cvClearMemStorage(storage)
cv.cvEqualizeHist(grayscale, grayscale)
cascade = cv.cvLoadHaarClassifierCascade(haar_file, cv.cvSize(1, 1))
objects = cv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2,
cv.CV_HAAR_DO_CANNY_PRUNING,
cv.cvSize(100, 100))
# Draw dots where hands are
if objects:
for i in objects:
#cv.cvRectangle(image, cv.cvPoint( int(i.x), int(i.y)),
# cv.cvPoint(int(i.x+i.width), int(i.y+i.height)),
# cv.CV_RGB(0,255,0), 3, 8, 0)
center = cv.cvPoint(int(i.x + i.width / 2),
int(i.y + i.height / 2))
cv.cvCircle(image, center, 10, cv.CV_RGB(0, 0, 0), 5, 8, 0)
# Left side check
if center.x > box_forward_left[
0].x and center.x < box_backwards_left[
1].x and center.y > box_forward_left[
0].y and center.y < box_backwards_left[1].y:
set_speed('left', center)
# Right side check
if center.x > box_forward_right[
0].x and center.x < box_backwards_right[
1].x and center.y > box_forward_right[
0].y and center.y < box_backwards_right[1].y:
set_speed('right', center)
def detect_circles(self, img_grey):
""" Find circles within the video stream """
cv.cvClearMemStorage(self.circles_storage)
circles = cv.cvHoughCircles(img_grey, self.circles_storage,
cv.CV_HOUGH_GRADIENT, 2,
img_grey.height / 4, 200, 100)
return circles
def detect_lines(self, img_grey, img_orig):
""" Detect lines within the image. To switch between standard and
probabilistic Hough transform, use cv.CV_HOUGH_STANDARD, or
cv.CV_HOUGH_PROBABILISTIC.
"""
# Set transform method ('standard','probabilistic')
transform_method = 'probabilistic'
# Clear out our storage
cv.cvClearMemStorage(self.lines_storage)
sz = cv.cvSize(img_grey.width & -2, img_grey.height & -2)
img_dst_color = cv.cvCreateImage(cv.cvGetSize(img_orig), 8, 3)
tgrey = cv.cvCreateImage(sz, 8, 1)
cv.cvCanny(tgrey, img_grey, 50, 200, 3)
if transform_method == 'standard':
lines = cv.cvHoughLines2(img_grey, self.lines_storage,
cv.CV_HOUGH_STANDARD, 1, cv.CV_PI / 180,
100, 0, 0)
else:
lines = cv.cvHoughLines2(img_grey, self.lines_storage,
cv.CV_HOUGH_PROBABILISTIC, 1,
cv.CV_PI / 180, 50, 50, 10)
return lines
def detect_lines(self, img_grey, img_orig):
""" Detect lines within the image. To switch between standard and
probabilistic Hough transform, use cv.CV_HOUGH_STANDARD, or
cv.CV_HOUGH_PROBABILISTIC.
"""
# Set transform method ('standard','probabilistic')
transform_method = 'probabilistic'
# Clear out our storage
cv.cvClearMemStorage(self.lines_storage)
sz = cv.cvSize(img_grey.width & -2, img_grey.height & -2)
img_dst_color = cv.cvCreateImage(cv.cvGetSize(img_orig), 8, 3)
tgrey = cv.cvCreateImage(sz, 8, 1)
cv.cvCanny(tgrey, img_grey, 50, 200, 3)
if transform_method == 'standard':
lines = cv.cvHoughLines2(img_grey,
self.lines_storage,
cv.CV_HOUGH_STANDARD,
1,
cv.CV_PI/180,
100,
0,
0)
else:
lines = cv.cvHoughLines2(img_grey,
self.lines_storage,
cv.CV_HOUGH_PROBABILISTIC,
1,
cv.CV_PI/180,
50,
50,
10)
return lines
def detect(image, cascade_file='haarcascade_data/haarcascade_frontalface_alt.xml'):
image_size = cv.cvGetSize(image)
# create grayscale version
grayscale = cv.cvCreateImage(image_size, 8, 1)
cv.cvCvtColor(image, grayscale, cv.CV_BGR2GRAY)
# create storage
storage = cv.cvCreateMemStorage(0)
cv.cvClearMemStorage(storage)
# equalize histogram
cv.cvEqualizeHist(grayscale, grayscale)
# detect objects
cascade = cv.cvLoadHaarClassifierCascade(cascade_file, cv.cvSize(1,1))
faces = cv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2, cv.CV_HAAR_DO_CANNY_PRUNING, cv.cvSize(50, 50))
positions = []
if faces:
for i in faces:
positions.append({'x': i.x, 'y': i.y, 'width': i.width, 'height':
i.height,})
cv.cvRectangle(image, cv.cvPoint( int(i.x), int(i.y)),
cv.cvPoint(int(i.x + i.width), int(i.y + i.height)),
cv.CV_RGB(0, 255, 0), 3, 8, 0)
return positions
def detect_faces(self, img_grey):
""" Detect faces within an image, then draw around them.
The default parameters (scale_factor=1.1, min_neighbors=3, flags=0) are tuned
for accurate yet slow object detection. For a faster operation on real video
images the settings are:
scale_factor=1.2, min_neighbors=2, flags=CV_HAAR_DO_CANNY_PRUNING,
min_size=<minimum possible face size
"""
min_size = cv.cvSize(20, 20)
self.image_scale = 1.3
haar_scale = 1.2
min_neighbors = 2
haar_flags = 0
# Create a small image for better performance
small_size = cv.cvSize(cv.cvRound(img_grey.width / self.image_scale),
cv.cvRound(img_grey.height / self.image_scale))
small_img = cv.cvCreateImage(small_size, 8, 1)
cv.cvResize(img_grey, small_img, cv.CV_INTER_LINEAR)
cv.cvEqualizeHist(small_img, small_img)
cv.cvClearMemStorage(self.faces_storage)
if (self.cascade):
t = cv.cvGetTickCount()
faces = cv.cvHaarDetectObjects(small_img, self.cascade,
self.faces_storage, haar_scale,
min_neighbors, haar_flags, min_size)
t = cv.cvGetTickCount() - t
cv.cvReleaseImage(small_img)
#print "detection time = %gms" % (t/(cvGetTickFrequency()*1000.));
return faces
def _detect(image):
""" Detects faces on `image`
Parameters:
@image: image file path
Returns:
[((x1, y1), (x2, y2)), ...] List of coordenates for top-left
and bottom-right corner
"""
# the OpenCV API says this function is obsolete, but we can't
# cast the output of cvLoad to a HaarClassifierCascade, so use
# this anyways the size parameter is ignored
capture = cvCreateFileCapture(image)
if not capture:
return []
frame = cvQueryFrame(capture)
if not frame:
return []
img = cvCreateImage(cvSize(frame.width, frame.height),
IPL_DEPTH_8U, frame.nChannels)
cvCopy(frame, img)
# allocate temporary images
gray = cvCreateImage((img.width, img.height),
COPY_DEPTH, COPY_CHANNELS)
width, height = (cvRound(img.width / IMAGE_SCALE),
cvRound(img.height / IMAGE_SCALE))
small_img = cvCreateImage((width, height), COPY_DEPTH, COPY_CHANNELS)
# convert color input image to grayscale
cvCvtColor(img, gray, CV_BGR2GRAY)
# scale input image for faster processing
cvResize(gray, small_img, CV_INTER_LINEAR)
cvEqualizeHist(small_img, small_img)
cvClearMemStorage(STORAGE)
coords = []
for haar_file in CASCADES:
cascade = cvLoadHaarClassifierCascade(haar_file, cvSize(1, 1))
if cascade:
faces = cvHaarDetectObjects(small_img, cascade, STORAGE, HAAR_SCALE,
MIN_NEIGHBORS, HAAR_FLAGS, MIN_SIZE) or []
for face_rect in faces:
# the input to cvHaarDetectObjects was resized, so scale the
# bounding box of each face and convert it to two CvPoints
x, y = face_rect.x, face_rect.y
pt1 = (int(x * IMAGE_SCALE), int(y * IMAGE_SCALE))
pt2 = (int((x + face_rect.width) * IMAGE_SCALE),
int((y + face_rect.height) * IMAGE_SCALE))
coords.append((pt1, pt2))
return coords
def _detect(image):
""" Detects faces on `image`
Parameters:
@image: image file path
Returns:
[((x1, y1), (x2, y2)), ...] List of coordenates for top-left
and bottom-right corner
"""
# the OpenCV API says this function is obsolete, but we can't
# cast the output of cvLoad to a HaarClassifierCascade, so use
# this anyways the size parameter is ignored
capture = cvCreateFileCapture(image)
if not capture:
return []
frame = cvQueryFrame(capture)
if not frame:
return []
img = cvCreateImage(cvSize(frame.width, frame.height), IPL_DEPTH_8U,
frame.nChannels)
cvCopy(frame, img)
# allocate temporary images
gray = cvCreateImage((img.width, img.height), COPY_DEPTH, COPY_CHANNELS)
width, height = (cvRound(img.width / IMAGE_SCALE),
cvRound(img.height / IMAGE_SCALE))
small_img = cvCreateImage((width, height), COPY_DEPTH, COPY_CHANNELS)
# convert color input image to grayscale
cvCvtColor(img, gray, CV_BGR2GRAY)
# scale input image for faster processing
cvResize(gray, small_img, CV_INTER_LINEAR)
cvEqualizeHist(small_img, small_img)
cvClearMemStorage(STORAGE)
coords = []
for haar_file in CASCADES:
cascade = cvLoadHaarClassifierCascade(haar_file, cvSize(1, 1))
if cascade:
faces = cvHaarDetectObjects(small_img, cascade, STORAGE,
HAAR_SCALE, MIN_NEIGHBORS, HAAR_FLAGS,
MIN_SIZE) or []
for face_rect in faces:
# the input to cvHaarDetectObjects was resized, so scale the
# bounding box of each face and convert it to two CvPoints
x, y = face_rect.x, face_rect.y
pt1 = (int(x * IMAGE_SCALE), int(y * IMAGE_SCALE))
pt2 = (int((x + face_rect.width) * IMAGE_SCALE),
int((y + face_rect.height) * IMAGE_SCALE))
coords.append((pt1, pt2))
return coords
def detect_circles(self, img_grey): """ Find circles within the video stream """ cv.cvClearMemStorage(self.circles_storage) circles = cv.cvHoughCircles(img_grey, self.circles_storage, cv.CV_HOUGH_GRADIENT, 2, img_grey.height/4, 200, 100) return circles
def detect_faces_on(path):
faces = []
image = cvLoadImage(path)
# convert to grayscale for faster results
grayscale = cvCreateImage(cvSize(image.width, image.height), 8, 1)
cvCvtColor(image, grayscale, CV_BGR2GRAY)
# smooth picture for better results
cvSmooth(grayscale, grayscale, CV_GAUSSIAN, 3, 3)
storage = cvCreateMemStorage(0)
cvClearMemStorage(storage)
cvEqualizeHist(grayscale, grayscale)
cascade_files = [
# ('/usr/share/opencv/haarcascades/haarcascade_eye_tree_eyeglasses.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_frontalface_alt.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_lowerbody.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_mouth.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_profileface.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_eye.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_frontalface_default.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_eyepair_big.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_nose.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_righteye_2splits.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_frontalface_alt2.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_fullbody.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_eyepair_small.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_righteye.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_upperbody.xml', (50, 50)),
('/usr/share/opencv/haarcascades/haarcascade_frontalface_alt_tree.xml',
(50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_lefteye_2splits.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_lefteye.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_upperbody.xml', (50, 50)),
# ('parojos_22_5.1.xml', (22, 5)),
# ('Mouth.xml', (22, 15)),
]
for cascade_file, cascade_sizes in cascade_files:
cascade = cvLoadHaarClassifierCascade(os.path.join(cascade_file),
cvSize(1, 1))
faces += cvHaarDetectObjects(grayscale, cascade, storage, HAAR_SCALE,
HAAR_NEIGHBORS, CV_HAAR_DO_CANNY_PRUNING,
cvSize(*cascade_sizes))
return [{'x': f.x, 'y': f.y, 'w': f.width, 'h': f.height} for f in faces]
def detect_faces(image):
"""Converts an image to grayscale and prints the locations of any
faces found"""
grayscale = cvCreateImage(cvSize(image.width, image.height), 8, 1)
cvCvtColor(image, grayscale, CV_BGR2GRAY)
storage = cvCreateMemStorage(0)
cvClearMemStorage(storage)
cvEqualizeHist(grayscale, grayscale)
# The default parameters (scale_factor=1.1, min_neighbors=3,
# flags=0) are tuned for accurate yet slow face detection. For
# faster face detection on real video images the better settings are
# (scale_factor=1.2, min_neighbors=2, flags=CV_HAAR_DO_CANNY_PRUNING).
# --- http://www710.univ-lyon1.fr/~bouakaz/OpenCV-0.9.5/docs/ref/OpenCVRef_Experimental.htm#decl_cvHaarDetectObjects
# The size box is of the *minimum* detectable object size. Smaller box = more processing time. - http://cell.fixstars.com/opencv/index.php/Facedetect
minsize = (int(MINFACEWIDTH_PERCENT * image.width + 0.5),
int(MINFACEHEIGHT_PERCENT * image.height))
print >> sys.stderr, "Min size of face: %s" % ` minsize `
faces = []
for cascadefile in [
'/usr/share/opencv/haarcascades/haarcascade_frontalface_default.xml'
]:
# for cascadefile in ['/usr/share/opencv/haarcascades/haarcascade_frontalface_default.xml', '/usr/share/opencv/haarcascades/haarcascade_profileface.xml']:
cascade = cvLoadHaarClassifierCascade(cascadefile, cvSize(1, 1))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2, CV_HAAR_DO_CANNY_PRUNING, cvSize(50,50))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1, 3, 0, cvSize(MINFACEWIDTH,MINFACEHEIGHT))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1, 3, 0, cvSize(MINFACEWIDTH,MINFACEHEIGHT))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1, 3, CV_HAAR_DO_CANNY_PRUNING, cvSize(*minsize))
faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1,
4, CV_HAAR_DO_CANNY_PRUNING,
cvSize(*minsize))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, scale_factor=1.1, min_neighbors=3, flags=0, cvSize(50,50))
# print dir(faces)
bboxes = []
if faces:
for f in faces:
print >> sys.stderr, "\tFace at [(%d,%d) -> (%d,%d)]" % (
f.x, f.y, f.x + f.width, f.y + f.height)
bboxes = [Face(f.x, f.y, f.x + f.width, f.y + f.height) for f in faces]
return bboxes
def detect(image):
image_size = opencv.cvGetSize(image)
# create grayscale version
grayscale = opencv.cvCreateImage(image_size, 8, 1)
opencv.cvCvtColor(image, grayscale, opencv.CV_BGR2GRAY)
# create storage
storage = opencv.cvCreateMemStorage(0)
opencv.cvClearMemStorage(storage)
# equalize histogram
opencv.cvEqualizeHist(grayscale, grayscale)
# detect objects
faces = opencv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2, opencv.CV_HAAR_DO_CANNY_PRUNING, opencv.cvSize(100, 100))
# eyes = opencv.cvHaarDetectObjects(grayscale, eye_cascade, storage, 1.2, 2, opencv.CV_HAAR_DO_CANNY_PRUNING, opencv.cvSize(60,60))
draw_bounding_boxes(faces, image, 127,255,0, 3)
def detectObjects(image):
"""Converts an image to grayscale and prints the locations of any
faces found"""
size = cv.cvSize(image.width, image.height)
grayscale = cv.cvCreateImage(size, 8, 1)
cv.cvCvtColor(image, grayscale, cv.CV_BGR2GRAY)
storage = cv.cvCreateMemStorage(0)
cv.cvClearMemStorage(storage)
cv.cvEqualizeHist(grayscale, grayscale)
cascade = cv.cvLoadHaarClassifierCascade(PATH, cv.cvSize(1,1))
faces = cv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2, cv.CV_HAAR_DO_CANNY_PRUNING, cv.cvSize(30,30))
if faces:
for f in faces:
print("[(%d,%d) -> (%d,%d)]" % (f.x, f.y, f.x+f.width, f.y+f.height))
def detect_faces_on(path):
faces = []
image = cvLoadImage(path)
# convert to grayscale for faster results
grayscale = cvCreateImage(cvSize(image.width, image.height), 8, 1)
cvCvtColor(image, grayscale, CV_BGR2GRAY)
# smooth picture for better results
cvSmooth(grayscale, grayscale, CV_GAUSSIAN, 3, 3)
storage = cvCreateMemStorage(0)
cvClearMemStorage(storage)
cvEqualizeHist(grayscale, grayscale)
cascade_files = [
# ('/usr/share/opencv/haarcascades/haarcascade_eye_tree_eyeglasses.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_frontalface_alt.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_lowerbody.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_mouth.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_profileface.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_eye.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_frontalface_default.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_eyepair_big.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_nose.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_righteye_2splits.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_frontalface_alt2.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_fullbody.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_eyepair_small.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_righteye.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_upperbody.xml', (50, 50)),
('/usr/share/opencv/haarcascades/haarcascade_frontalface_alt_tree.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_lefteye_2splits.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_lefteye.xml', (50, 50)),
# ('/usr/share/opencv/haarcascades/haarcascade_mcs_upperbody.xml', (50, 50)),
# ('parojos_22_5.1.xml', (22, 5)),
# ('Mouth.xml', (22, 15)),
]
for cascade_file, cascade_sizes in cascade_files:
cascade = cvLoadHaarClassifierCascade(os.path.join(cascade_file), cvSize(1, 1))
faces += cvHaarDetectObjects(grayscale, cascade, storage, HAAR_SCALE, HAAR_NEIGHBORS, CV_HAAR_DO_CANNY_PRUNING, cvSize(*cascade_sizes))
return [{'x': f.x, 'y': f.y, 'w': f.width, 'h': f.height} for f in faces]
def detect_face(self, img): """ Detect faces within an image, then draw around them. The default parameters (scale_factor=1.1, min_neighbors=3, flags=0) are tuned for accurate yet slow object detection. For a faster operation on real video images the settings are: scale_factor=1.2, min_neighbors=2, flags=CV_HAAR_DO_CANNY_PRUNING, min_size=<minimum possible face size """ min_size = cv.cvSize(20,20) image_scale = 1.3 haar_scale = 1.2 min_neighbors = 2 haar_flags = 0 gray = cv.cvCreateImage(cv.cvSize(img.width,img.height), 8, 1) small_img = cv.cvCreateImage(cv.cvSize(cv.cvRound(img.width/image_scale), cv.cvRound(img.height/image_scale)), 8, 1) cv.cvCvtColor(img, gray, cv.CV_BGR2GRAY) cv.cvResize(gray, small_img, cv.CV_INTER_LINEAR) cv.cvEqualizeHist(small_img, small_img) cv.cvClearMemStorage(self.storage) if(self.cascade): t = cv.cvGetTickCount(); faces = cv.cvHaarDetectObjects(small_img, self.cascade, self.storage, haar_scale, min_neighbors, haar_flags, min_size) t = cv.cvGetTickCount() - t #print "detection time = %gms" % (t/(cvGetTickFrequency()*1000.)); if faces: for r in faces: pt1 = cv.cvPoint(int(r.x*image_scale), int(r.y*image_scale)) pt2 = cv.cvPoint(int((r.x+r.width)*image_scale), int((r.y+r.height)*image_scale)) cv.cvRectangle(img, pt1, pt2, cv.CV_RGB(255,0,0), 3, 8, 0) return img
def detect(image):
image_size = cv.cvGetSize(image)
# create grayscale version
grayscale = cv.cvCreateImage(image_size, 8, 1)
cv.cvCvtColor(image, grayscale, opencv.CV_BGR2GRAY)
# create storage
storage = cv.cvCreateMemStorage(0)
cv.cvClearMemStorage(storage)
# equalize histogram
cv.cvEqualizeHist(grayscale, grayscale)
# detect objects
cascade = cv.cvLoadHaarClassifierCascade('haarcascade_frontalface_alt.xml', cv.cvSize(1,1))
faces = cv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2, opencv.CV_HAAR_DO_CANNY_PRUNING, cv.cvSize(100, 100))
if faces:
for i in faces:
r = image[int(i.y):int(i.y+i.height),int(i.x):int(i.x+i.width)]
cv.cvSmooth(r,r,cv.CV_BLUR,51,51)
def detect(self, pil_image, cascade_name, recogn_w = 50, recogn_h = 50):
# Get cascade:
cascade = self.get_cascade(cascade_name)
image = opencv.PIL2Ipl(pil_image)
image_size = opencv.cvGetSize(image)
grayscale = image
if pil_image.mode == "RGB":
# create grayscale version
grayscale = opencv.cvCreateImage(image_size, 8, 1)
# Change to RGB2Gray - I dont think itll affect the conversion
opencv.cvCvtColor(image, grayscale, opencv.CV_BGR2GRAY)
# create storage
storage = opencv.cvCreateMemStorage(0)
opencv.cvClearMemStorage(storage)
# equalize histogram
opencv.cvEqualizeHist(grayscale, grayscale)
# detect objects
return opencv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2, opencv.CV_HAAR_DO_CANNY_PRUNING, opencv.cvSize(recogn_w, recogn_h))
def detect_face(self, img):
""" Detect faces within an image, then draw around them.
The default parameters (scale_factor=1.1, min_neighbors=3, flags=0) are tuned
for accurate yet slow object detection. For a faster operation on real video
images the settings are:
scale_factor=1.2, min_neighbors=2, flags=CV_HAAR_DO_CANNY_PRUNING,
min_size=<minimum possible face size
"""
min_size = cv.cvSize(20, 20)
image_scale = 1.3
haar_scale = 1.2
min_neighbors = 2
haar_flags = 0
gray = cv.cvCreateImage(cv.cvSize(img.width, img.height), 8, 1)
small_img = cv.cvCreateImage(
cv.cvSize(cv.cvRound(img.width / image_scale),
cv.cvRound(img.height / image_scale)), 8, 1)
cv.cvCvtColor(img, gray, cv.CV_BGR2GRAY)
cv.cvResize(gray, small_img, cv.CV_INTER_LINEAR)
cv.cvEqualizeHist(small_img, small_img)
cv.cvClearMemStorage(self.storage)
if (self.cascade):
t = cv.cvGetTickCount()
faces = cv.cvHaarDetectObjects(small_img, self.cascade,
self.storage, haar_scale,
min_neighbors, haar_flags, min_size)
t = cv.cvGetTickCount() - t
#print "detection time = %gms" % (t/(cvGetTickFrequency()*1000.));
if faces:
for r in faces:
pt1 = cv.cvPoint(int(r.x * image_scale),
int(r.y * image_scale))
pt2 = cv.cvPoint(int((r.x + r.width) * image_scale),
int((r.y + r.height) * image_scale))
cv.cvRectangle(img, pt1, pt2, cv.CV_RGB(255, 0, 0), 3, 8,
0)
return img
def detect_faces(image):
"""Converts an image to grayscale and prints the locations of any
faces found"""
grayscale = cvCreateImage(cvSize(image.width, image.height), 8, 1)
cvCvtColor(image, grayscale, CV_BGR2GRAY)
storage = cvCreateMemStorage(0)
cvClearMemStorage(storage)
cvEqualizeHist(grayscale, grayscale)
# The default parameters (scale_factor=1.1, min_neighbors=3,
# flags=0) are tuned for accurate yet slow face detection. For
# faster face detection on real video images the better settings are
# (scale_factor=1.2, min_neighbors=2, flags=CV_HAAR_DO_CANNY_PRUNING).
# --- http://www710.univ-lyon1.fr/~bouakaz/OpenCV-0.9.5/docs/ref/OpenCVRef_Experimental.htm#decl_cvHaarDetectObjects
# The size box is of the *minimum* detectable object size. Smaller box = more processing time. - http://cell.fixstars.com/opencv/index.php/Facedetect
minsize = (int(MINFACEWIDTH_PERCENT*image.width+0.5),int(MINFACEHEIGHT_PERCENT*image.height))
print >> sys.stderr, "Min size of face: %s" % `minsize`
faces = []
for cascadefile in ['/usr/share/opencv/haarcascades/haarcascade_frontalface_default.xml']:
# for cascadefile in ['/usr/share/opencv/haarcascades/haarcascade_frontalface_default.xml', '/usr/share/opencv/haarcascades/haarcascade_profileface.xml']:
cascade = cvLoadHaarClassifierCascade(cascadefile, cvSize(1,1))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2, CV_HAAR_DO_CANNY_PRUNING, cvSize(50,50))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1, 3, 0, cvSize(MINFACEWIDTH,MINFACEHEIGHT))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1, 3, 0, cvSize(MINFACEWIDTH,MINFACEHEIGHT))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1, 3, CV_HAAR_DO_CANNY_PRUNING, cvSize(*minsize))
faces += cvHaarDetectObjects(grayscale, cascade, storage, 1.1, 4, CV_HAAR_DO_CANNY_PRUNING, cvSize(*minsize))
# faces += cvHaarDetectObjects(grayscale, cascade, storage, scale_factor=1.1, min_neighbors=3, flags=0, cvSize(50,50))
# print dir(faces)
bboxes = []
if faces:
for f in faces:
print >> sys.stderr, "\tFace at [(%d,%d) -> (%d,%d)]" % (f.x, f.y, f.x+f.width, f.y+f.height)
bboxes = [Face(f.x, f.y, f.x+f.width, f.y+f.height) for f in faces]
return bboxes
def detect(self, pil_image, cascade_name, recogn_w=50, recogn_h=50):
# Get cascade:
cascade = self.get_cascade(cascade_name)
image = opencv.PIL2Ipl(pil_image)
image_size = opencv.cvGetSize(image)
grayscale = image
if pil_image.mode == "RGB":
# create grayscale version
grayscale = opencv.cvCreateImage(image_size, 8, 1)
# Change to RGB2Gray - I dont think itll affect the conversion
opencv.cvCvtColor(image, grayscale, opencv.CV_BGR2GRAY)
# create storage
storage = opencv.cvCreateMemStorage(0)
opencv.cvClearMemStorage(storage)
# equalize histogram
opencv.cvEqualizeHist(grayscale, grayscale)
# detect objects
return opencv.cvHaarDetectObjects(grayscale, cascade, storage, 1.2, 2,
opencv.CV_HAAR_DO_CANNY_PRUNING,
opencv.cvSize(recogn_w, recogn_h))
def detect_squares(self, img_grey, img_orig): """ Find squares within the video stream and draw them """ cv.cvClearMemStorage(self.faces_storage) N = 11 thresh = 5 sz = cv.cvSize(img_grey.width & -2, img_grey.height & -2) timg = cv.cvCloneImage(img_orig) pyr = cv.cvCreateImage(cv.cvSize(sz.width/2, sz.height/2), 8, 3) # create empty sequence that will contain points - # 4 points per square (the square's vertices) squares = cv.cvCreateSeq(0, cv.sizeof_CvSeq, cv.sizeof_CvPoint, self.squares_storage) squares = cv.CvSeq_CvPoint.cast(squares) # select the maximum ROI in the image # with the width and height divisible by 2 subimage = cv.cvGetSubRect(timg, cv.cvRect(0, 0, sz.width, sz.height)) cv.cvReleaseImage(timg) # down-scale and upscale the image to filter out the noise cv.cvPyrDown(subimage, pyr, 7) cv.cvPyrUp(pyr, subimage, 7) cv.cvReleaseImage(pyr) tgrey = cv.cvCreateImage(sz, 8, 1) # find squares in every color plane of the image for c in range(3): # extract the c-th color plane channels = [None, None, None] channels[c] = tgrey cv.cvSplit(subimage, channels[0], channels[1], channels[2], None) for l in range(N): # hack: use Canny instead of zero threshold level. # Canny helps to catch squares with gradient shading if(l == 0): # apply Canny. Take the upper threshold from slider # and set the lower to 0 (which forces edges merging) cv.cvCanny(tgrey, img_grey, 0, thresh, 5) # dilate canny output to remove potential # holes between edge segments cv.cvDilate(img_grey, img_grey, None, 1) else: # apply threshold if l!=0: # tgray(x,y) = gray(x,y) < (l+1)*255/N ? 255 : 0 cv.cvThreshold(tgrey, img_grey, (l+1)*255/N, 255, cv.CV_THRESH_BINARY) # find contours and store them all as a list count, contours = cv.cvFindContours(img_grey, self.squares_storage, cv.sizeof_CvContour, cv.CV_RETR_LIST, cv.CV_CHAIN_APPROX_SIMPLE, cv.cvPoint(0,0)) if not contours: continue # test each contour for contour in contours.hrange(): # approximate contour with accuracy proportional # to the contour perimeter result = cv.cvApproxPoly(contour, cv.sizeof_CvContour, self.squares_storage, cv.CV_POLY_APPROX_DP, cv.cvContourPerimeter(contours)*0.02, 0) # square contours should have 4 vertices after approximation # relatively large area (to filter out noisy contours) # and be convex. # Note: absolute value of an area is used because # area may be positive or negative - in accordance with the # contour orientation if(result.total == 4 and abs(cv.cvContourArea(result)) > 1000 and cv.cvCheckContourConvexity(result)): s = 0 for i in range(5): # find minimum angle between joint # edges (maximum of cosine) if(i >= 2): t = abs(self.squares_angle(result[i], result[i-2], result[i-1])) if s<t: s = t # if cosines of all angles are small # (all angles are ~90 degree) then write quandrange # vertices to resultant sequence if(s < 0.3): for i in range(4): squares.append(result[i]) cv.cvReleaseImage(tgrey) return squares
def detect_squares(self, img_grey, img_orig):
""" Find squares within the video stream and draw them """
cv.cvClearMemStorage(self.faces_storage)
N = 11
thresh = 5
sz = cv.cvSize(img_grey.width & -2, img_grey.height & -2)
timg = cv.cvCloneImage(img_orig)
pyr = cv.cvCreateImage(cv.cvSize(sz.width / 2, sz.height / 2), 8, 3)
# create empty sequence that will contain points -
# 4 points per square (the square's vertices)
squares = cv.cvCreateSeq(0, cv.sizeof_CvSeq, cv.sizeof_CvPoint,
self.squares_storage)
squares = cv.CvSeq_CvPoint.cast(squares)
# select the maximum ROI in the image
# with the width and height divisible by 2
subimage = cv.cvGetSubRect(timg, cv.cvRect(0, 0, sz.width, sz.height))
cv.cvReleaseImage(timg)
# down-scale and upscale the image to filter out the noise
cv.cvPyrDown(subimage, pyr, 7)
cv.cvPyrUp(pyr, subimage, 7)
cv.cvReleaseImage(pyr)
tgrey = cv.cvCreateImage(sz, 8, 1)
# find squares in every color plane of the image
for c in range(3):
# extract the c-th color plane
channels = [None, None, None]
channels[c] = tgrey
cv.cvSplit(subimage, channels[0], channels[1], channels[2], None)
for l in range(N):
# hack: use Canny instead of zero threshold level.
# Canny helps to catch squares with gradient shading
if (l == 0):
# apply Canny. Take the upper threshold from slider
# and set the lower to 0 (which forces edges merging)
cv.cvCanny(tgrey, img_grey, 0, thresh, 5)
# dilate canny output to remove potential
# holes between edge segments
cv.cvDilate(img_grey, img_grey, None, 1)
else:
# apply threshold if l!=0:
# tgray(x,y) = gray(x,y) < (l+1)*255/N ? 255 : 0
cv.cvThreshold(tgrey, img_grey, (l + 1) * 255 / N, 255,
cv.CV_THRESH_BINARY)
# find contours and store them all as a list
count, contours = cv.cvFindContours(img_grey,
self.squares_storage,
cv.sizeof_CvContour,
cv.CV_RETR_LIST,
cv.CV_CHAIN_APPROX_SIMPLE,
cv.cvPoint(0, 0))
if not contours:
continue
# test each contour
for contour in contours.hrange():
# approximate contour with accuracy proportional
# to the contour perimeter
result = cv.cvApproxPoly(
contour, cv.sizeof_CvContour, self.squares_storage,
cv.CV_POLY_APPROX_DP,
cv.cvContourPerimeter(contours) * 0.02, 0)
# square contours should have 4 vertices after approximation
# relatively large area (to filter out noisy contours)
# and be convex.
# Note: absolute value of an area is used because
# area may be positive or negative - in accordance with the
# contour orientation
if (result.total == 4
and abs(cv.cvContourArea(result)) > 1000
and cv.cvCheckContourConvexity(result)):
s = 0
for i in range(5):
# find minimum angle between joint
# edges (maximum of cosine)
if (i >= 2):
t = abs(
self.squares_angle(result[i],
result[i - 2],
result[i - 1]))
if s < t:
s = t
# if cosines of all angles are small
# (all angles are ~90 degree) then write quandrange
# vertices to resultant sequence
if (s < 0.3):
for i in range(4):
squares.append(result[i])
cv.cvReleaseImage(tgrey)
return squares
