def detect(self,img,face_records,options):
'''Run a face detector and return rectangles.'''
# TODO: Make this an option
detection_threshold = options.threshold
if options.best:
detection_threshold = 0.01
#thresh=.5
#hier_thresh=.5
#nms=.45
thresh=self.t1
hier_thresh=self.t2
#nms=self.t_nms
start = time.time()
# Reformat the image for darknet
img = (img/255.0).astype(np.float32)
img = np.transpose(img,(2,0,1)).copy()
c,h,w = img.shape
im = IMAGE(w,h,c,img.ctypes.data_as(ctypes.POINTER(ctypes.c_float)))
num = c_int(0)
pnum = pointer(num)
self.predict_image(self.net, im)
dets = self.get_network_boxes(self.net, im.w, im.h, thresh, hier_thresh, None, 0, pnum)
num = pnum[0]
if (detection_threshold > 0.0): self.do_nms_obj(dets, num, self.meta.classes, detection_threshold);
# print('time checkBB:',time.time()-start)
res = []
for j in range(num):
for i in range(self.meta.classes):
if dets[j].prob[i] > 0:
b = dets[j].bbox
res.append((self.meta.names[i], dets[j].prob[i], (b.x, b.y, b.w, b.h)))
res = sorted(res, key=lambda x: -x[1])
#self.free_image(im)
self.free_detections(dets, num)
k = 0
for each in res:
name,prob,rect = each
face_record = face_records.face_records.add()
face_record.detection.score = prob
face_record.detection.location.CopyFrom(pt.rect_val2proto(rect[0]-0.5*rect[2],rect[1]-0.5*rect[3],rect[2],rect[3]))
face_record.detection.detection_id = k
face_record.detection.detection_class = name
k += 1
return
def detect(self, img, face_records, options):
'''Run a face detector and return rectangles.'''
detection_threshold = options.threshold
print('Detection Threshold', detection_threshold)
# TODO: Make this an option
if options.best:
detection_threshold = -1.5
# Run the detector on the image
dets = self.detector(img, 1)
# Now process each face we found and add a face to the records list.
for k, d in enumerate(dets):
face_record = face_records.face_records.add()
face_record.detection.score = d.confidence
face_record.detection.location.CopyFrom(
pt.rect_val2proto(d.rect.left(), d.rect.top(), d.rect.width(),
d.rect.height()))
face_record.detection.detection_id = k
face_record.detection.detection_class = "FACE_%d" % k
shape = self.shape_pred(img, d.rect)
for i in range(len(shape.parts())):
loc = shape.parts()[i]
landmark = face_record.landmarks.add()
landmark.landmark_id = "point_%02d" % i
landmark.location.x = loc.x
landmark.location.y = loc.y
if options.best:
face_records.face_records.sort(key=lambda x: -x.detection.score)
while len(face_records.face_records) > 1:
del face_records.face_records[-1]
def filterDetectBest(face_records, im, best):
# TODO: This is a temporary fix.
if best and len(face_records.face_records) > 1:
print(
"WARNING: detector service does not seem to support best mode. Too many faces returned."
)
face_records.face_records.sort(key=lambda x: -x.detection.score)
#print(detections.detections)
while len(face_records.face_records) > 1:
del face_records.face_records[-1]
assert len(face_records.face_records) == 1
if best and len(face_records.face_records) == 0:
print(
"WARNING: detector service does not seem to support best mode. No faces returned."
)
# in this case select the center of the image
det = face_records.face_records.add().detection
h, w = im.shape[:2]
s = 0.8 * min(w, h)
det.location.CopyFrom(
pt.rect_val2proto(0.5 * w - 0.5 * s, 0.5 * h - 0.5 * s, s, s))
det.score = -1.0
det.detection_id = 1
assert len(face_records.face_records) == 1
return face_records
def detect(self, img, face_records, options):
'''Run a face detector and return rectangles.'''
print('Running Face Detector For ArchFace')
#print(options.threshold)
# Run the detector on the image
dets, lpts = self.detector.detect(img,
threshold=options.threshold,
scale=1)
# Now process each face we found and add a face to the records list.
for idx in range(0, dets.shape[0]):
face_record = face_records.face_records.add()
face_record.detection.score = dets[idx, -1:]
ulx, uly, lrx, lry = dets[idx, :-1]
#create_square_bbox = np.amax(abs(lrx-ulx) , abs(lry-uly))
face_record.detection.location.CopyFrom(
pt.rect_val2proto(ulx, uly, abs(lrx - ulx), abs(lry - uly)))
face_record.detection.detection_id = idx
face_record.detection.detection_class = "FACE_%d" % idx
lmark = face_record.landmarks.add()
lmarkloc = lpts[idx]
for ldx in range(0, lmarkloc.shape[0]):
lmark.landmark_id = "point_%02d" % ldx
lmark.location.x = lmarkloc[ldx][0]
lmark.location.y = lmarkloc[ldx][1]
if options.best:
face_records.face_records.sort(key=lambda x: -x.detection.score)
while len(face_records.face_records) > 1:
del face_records.face_records[-1]
print('Done Running Face Detector For ArchFace')
def detect(self, img, face_records, options):
'''Run a face detector and return rectangles.'''
#print('Running Face Detector For ArchFace')
img = img[:, :, ::
-1] #convert from rgb to bgr . There is a reordering from bgr to RGB internally in the detector code.
dets, lpts = self.detector.detect(img,
threshold=options.threshold,
scale=1)
#print('Number of detections ', dets.shape[0])
# Now process each face we found and add a face to the records list.
for idx in range(0, dets.shape[0]):
face_record = face_records.face_records.add()
try: # todo: this seems to be different depending on mxnet version
face_record.detection.score = dets[idx, -1:]
except:
face_record.detection.score = dets[idx, -1:][0]
ulx, uly, lrx, lry = dets[idx, :-1]
#create_square_bbox = np.amax(abs(lrx-ulx) , abs(lry-uly))
face_record.detection.location.CopyFrom(
pt.rect_val2proto(ulx, uly, abs(lrx - ulx), abs(lry - uly)))
face_record.detection.detection_id = idx
face_record.detection.detection_class = "FACE_%d" % idx
#lmark = face_record.landmarks.add()
lmarkloc = lpts[idx]
for ldx in range(0, lmarkloc.shape[0]):
lmark = face_record.landmarks.add()
lmark.landmark_id = "point_%02d" % ldx
lmark.location.x = lmarkloc[ldx][0]
lmark.location.y = lmarkloc[ldx][1]
if options.best:
face_records.face_records.sort(key=lambda x: -x.detection.score)
while len(face_records.face_records) > 1:
del face_records.face_records[-1]
def detect(self,img,face_records,options):
'''Run a face detector and return rectangles.'''
# TODO: Make this an option
detection_threshold = options.threshold
if options.best:
detection_threshold = -1.5
# Run the detector on the image
dets, scores, idx = self.detector.run(img, 1, detection_threshold)
# Now process each face we found and add a face to the records list.
for k, d in enumerate(dets):
face_record = face_records.face_records.add()
face_record.detection.score = scores[k]
face_record.detection.location.CopyFrom(pt.rect_val2proto(d.left(), d.top(), d.width(), d.height()))
face_record.detection.detection_id = k
face_record.detection.detection_class = "FACE_%d"%idx[k]
if options.best:
face_records.face_records.sort(key = lambda x: -x.detection.score)
while len(face_records.face_records) > 1:
del face_records.face_records[-1]
def detect(self, img, face_records, options):
detected_templates = self._detect(img, options)
for i in range(0, self.num_faces):
curr_template = roc.roc_template_array_getitem(
detected_templates, i)
if curr_template.algorithm_id & roc.ROC_INVALID or curr_template.algorithm_id == 0:
continue
else:
face_record = face_records.face_records.add()
face_record.detection.score = curr_template.detection.confidence
xc, yc, w, h = curr_template.detection.x, curr_template.detection.y, curr_template.detection.width, curr_template.detection.height
x = int(xc - (w * 0.5))
y = int(yc - (w * 0.5))
face_record.detection.location.CopyFrom(
pt.rect_val2proto(x, y, w, h))
face_record.detection.detection_id = i
face_record.detection.detection_class = "FACE"
face_record.template.buffer = self._rocFlatten(curr_template)
#Free all the roc stuff
for i in range(0, self.num_faces):
roc.roc_free_template(
roc.roc_template_array_getitem(detected_templates, i))
def detect(self, img, face_records, options):
'''Run a face detector and return rectangles.'''
# Load the model
mat = img
#print('options<',options,'>')
#thresh = options.threshold
#result_id = 0
#im = mat #cv2.imread(pathname)
timer = pv.Timer()
# Run the network in the worker processes...
dets = self.runNetwork(mat, options)
#dets = as_result.get()
dets[:, 2] = dets[:, 2] - dets[:, 0] + 1
dets[:, 3] = dets[:, 3] - dets[:, 1] + 1
timer.mark("End Detection")
#print('dets')
# Now process each face we found and add a face to the records list.
for k, d in enumerate(dets):
face_record = face_records.face_records.add()
face_record.detection.score = d[4]
face_record.detection.location.CopyFrom(
pt.rect_val2proto(d[0], d[1], d[2], d[3]))
face_record.detection.detection_id = k
face_record.detection.detection_class = "FACE"
if options.best:
face_records.face_records.sort(key=lambda x: -x.detection.score)
while len(face_records.face_records) > 1:
del face_records.face_records[-1]
def detect(self, img, face_records, options):
'''Run a face detector and return rectangles.'''
#print('Running Face Detector For ArchFace')
img = img[:, :, ::
-1] #convert from rgb to bgr . There is a reordering from bgr to RGB internally in the detector code.
faces = self.app.get(img)
#print('Number of detections ', dets.shape[0])
# Now process each face we found and add a face to the records list.
idx = -1
for face in faces:
idx += 1
face_record = face_records.face_records.add()
face_record.detection.score = face.det_score
ulx, uly, lrx, lry = face.bbox
#create_square_bbox = np.amax(abs(lrx-ulx) , abs(lry-uly))
face_record.detection.location.CopyFrom(
pt.rect_val2proto(ulx, uly, abs(lrx - ulx), abs(lry - uly)))
face_record.detection.detection_id = idx
face_record.detection.detection_class = "FACE"
#lmark = face_record.landmarks.add()
for ldx in range(0, face.kps.shape[0]):
lmark = face_record.landmarks.add()
lmark.landmark_id = "point_%02d" % ldx
lmark.location.x = face.kps[ldx][0]
lmark.location.y = face.kps[ldx][1]
normed_embedding = face.normed_embedding
face_record.template.data.CopyFrom(
pt.vector_np2proto(normed_embedding))
if hasattr(face, 'landmark_2d_106'):
attribute = face_record.attributes.add()
attribute.key = 'landmark_2d_106'
attribute.matrix.CopyFrom(
pt.matrix_np2proto(face.landmark_2d_106))
if hasattr(face, 'landmark_3d_68'):
attribute = face_record.attributes.add()
attribute.key = 'landmark_3d_68'
attribute.matrix.CopyFrom(
pt.matrix_np2proto(face.landmark_3d_68))
if hasattr(face, 'sex'):
attribute = face_record.attributes.add()
attribute.key = 'sex'
attribute.text = face.sex
if hasattr(face, 'age'):
attribute = face_record.attributes.add()
attribute.key = 'age'
attribute.fvalue = face.age
if options.best:
face_records.face_records.sort(key=lambda x: -x.detection.score)
while len(face_records.face_records) > 1:
del face_records.face_records[-1]