def update_socket(ws):
print 'websocket connection request'
state['draw_output'] = True
while not ws.closed:
new_data_condition.acquire()
new_data_condition.wait()
new_data_condition.release()
result = {
'targets': state['targets'],
'fps': state['fps'],
'connected': state['ack']
}
_, binframe = cv2.imencode('.jpg', state['output_images']['bin'])
result['binaryImg'] = base64.b64encode(binframe)
_, binframe = cv2.imencode('.jpg', state['output_images']['result'])
result['resultImg'] = base64.b64encode(binframe)
message = json.dumps(result)
ws.send(message)
received = json.loads(ws.receive())
if 'thresholds' in received:
config['target'] = received['thresholds']
save_config(config)
if 'camera' in received:
config['camera'] = received['camera']
save_config(config)
print 'websocket disconnected'
state['draw_output'] = False
def get_frame(self,faced,saved=False, video=False, videoStop = False):
while True:
success, image = self.video.read()
if not success:
continue
else:
break
if saved:
cv2.imwrite("image.jpg",image)
if video:
self.videoWriter.write(image)
cv2.circle(image,(20,20), 15, (0,0,255), -1)
if faced:
faceCascade = cv2.CascadeClassifier("faceDetect.xml")
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
faces = faceCascade.detectMultiScale(
gray,
scaleFactor=1.1,
minNeighbors=5,
flags=cv2.cv.CV_HAAR_SCALE_IMAGE
)
for (x, y, w, h) in faces:
cv2.rectangle(image, (x, y), (x+w, y+h), (0, 255, 0), 2)
ret, jpeg = cv2.imencode('.jpg', image)
return (jpeg.tostring())
ret, jpeg = cv2.imencode('.jpg', image)
return (jpeg.tostring())
def getOrient():
global sock, capL, capR, frameL, frameR
retL, img_encodeL = cv2.imencode(
'.jpeg',
frameL[50:100],
encode_param
)
retR, img_encodeR = cv2.imencode(
'.jpeg',
frameR[50:100],
encode_param
)
stringDataL = np.array(img_encodeL).tostring()
stringDataR = np.array(img_encodeR).tostring()
client_commandL = 'put'.ljust(16)
if confirm(sock, client_commandL):
print client_commandL
sock.send(str(len(stringDataL)).ljust(16))
sock.send(stringDataL)
sock.send(str(len(stringDataR)).ljust(16))
sock.send(stringDataR)
else:
print 'server error!'
return False
orient = sock.recv(4096)
return int(orient)
def set_image(self, image, channel="default", storage_method=None):
"""
:param image: numpy array containing an RGB representation of an image
:type image: numpy.ndarray
:param channel: optional argument defining with which channel the image should be associated (default: "default")
:param channel: str
:return: Integer indicating whether the processing of the image was successful
:rtype: int
Inserts an image into a channel.
"""
ret = 0
if storage_method is None:
self._per_channel_storage_method[channel] = self._internal_storage_method
elif storage_method in ["raw", "png", "jpg"]:
self._per_channel_storage_method[channel] = storage_method
else:
raise Exception("Unknown storage method")
if self._per_channel_storage_method[channel] == "raw":
self._image_storage[self._channel[channel]] = image.copy()
ret = 0
if self._per_channel_storage_method[channel] == "png":
quality = 9 - int(self._compression_level/10)
(ret, buf) = cv2.imencode(".png", image, (cv2.IMWRITE_PNG_COMPRESSION, quality))
self._image_storage[self._channel[channel]] = buf
if self._per_channel_storage_method[channel] == "jpg":
(ret, buf) = cv2.imencode(".jpg", image, (cv2.IMWRITE_JPEG_QUALITY, self._compression_level))
self._image_storage[self._channel[channel]] = buf
return ret
def image(img, to_bgr=True, encoding='jpg', **kwargs):
""" image(img, [win, title, labels, width])
to_bgr: converts to bgr, if encoded as rgb (default True).
encoding: 'jpg' (default) or 'png'
"""
assert img.ndim == 2 or img.ndim == 3
win = kwargs.get('win') or uid()
if isinstance(img, list):
return images(img, kwargs)
# TODO: if img is a 3d tensor, then unstack it into a list of images
img = to_rgb(normalize(img, kwargs))
if to_bgr:
img = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
#pngbytes = png.encode(img.tostring(), img.shape[1], img.shape[0])
if encoding=='jpg':
jpgbytes = cv2.imencode('.jpg', img)[1]
imgdata = 'data:image/jpg;base64,' + base64.b64encode(jpgbytes).decode('ascii')
elif encoding=='png':
pngbytes = cv2.imencode('.png', img)[1]
imgdata = 'data:image/png;base64,' + base64.b64encode(pngbytes).decode('ascii')
else:
raise ValueError('unknown encoding')
send(command='image', id=win, src=imgdata,
labels=kwargs.get('labels'),
width=kwargs.get('width'),
title=kwargs.get('title'))
return win
def deal_img(imgFile):
try:
#img = cv2.imread(imgFile.encode('gbk'),1)
img = cv2.imdecode(np.fromfile(imgFile,dtype=np.uint8), -1)
x, y, z = img.shape # 高 宽 通道数
cant_deal_file_path = imgFile.replace("源文件", "未处理文件")
# if x < 600 and y < 1000: # 太小文件不处理
# shutil.copyfile(imgFile, cant_deal_file_path)
# print(datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S") + "," + imgFile + "文件太小，跳过")
# if y > 3500: # 太大的文件不处理
# shutil.copyfile(imgFile, cant_deal_file_path)
# print(datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S") + "," + imgFile + "文件太大，跳过")
if os.path.exists(imgFile): # 这就是最关键的代码了
print(imgFile)
for i in range(int(x/13)):
for j in range(int(y/3.15)+10):
varP = img[i, j]
img[i, j] = img[int(i/12.5), j]
# if sum(varP) > 556 and sum(varP) < 640: # 大于250，小于765（sum比白色的小）
# #img[i, j] = img[i - 5, j - 10]
#
# print(img[i, j])
dst = imgFile.replace("源文件", "目标文件")
#shutil.copyfile(imgFile, dst)
#cv2.imwrite(dst, img)
#cv2.imread(imgFile.decode('u8').encode('gbk'), -1)
cv2.imencode('.jpg', img)[1].tofile(dst)
#hutil.copyfile(imgFile, dst)
#shutil.copyfile(imgOldFile, imgOldFile)
except Exception as e:
shutil.copyfile(imgFile, cant_deal_file_path)
print("Exception:", e)
def measurePictures(self):
prvs = cv2.cvtColor(self.f1,cv2.COLOR_BGR2GRAY)
ret, self.frame1 = cv2.imencode('.png', self.f1)
ret = cv2.imwrite('frame1.png', self.f1)
next = cv2.cvtColor(self.f2,cv2.COLOR_BGR2GRAY)
flow = cv2.calcOpticalFlowFarneback(prvs,next, None, 0.5, 3, 15, 3, 5, 1.2, 0)
bgr = self.flow2hsv(flow)
(res, mx, my, ms) = self.flow2measure(flow)
self.showmeasure(bgr, res, mx, my, ms)
ret, self.flow12 = cv2.imencode('.png', bgr)
ret = cv2.imwrite('flow12.png', bgr)
ret, self.frame2 = cv2.imencode('.png', self.f2)
ret = cv2.imwrite('frame2.png', self.f2)
prvs = next
next = cv2.cvtColor(self.f3,cv2.COLOR_BGR2GRAY)
flow = cv2.calcOpticalFlowFarneback(prvs,next, None, 0.5, 3, 15, 3, 5, 1.2, 0)
bgr = self.flow2hsv(flow)
(res, mx, my, ms) = self.flow2measure(flow)
self.showmeasure(bgr, res, mx, my, ms)
ret, self.flow23 = cv2.imencode('.png', bgr)
ret = cv2.imwrite('flow23.png', bgr)
ret, self.frame3 = cv2.imencode('.png', self.f3)
ret = cv2.imwrite('frame3.png', self.f3)
def gen(value):
"""Video streaming generator function."""
soc = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
soc.settimeout(2)
dataSend = True
# connect to remote host
try:
soc.connect((host, port))
print 'found target'
soc.send('#Found you\n')
dataSend = True
except:
print("unable to connect")
dataSend = False
while 1:
if(dataSend == False):
try:
soc.connect((host,port))
print('found target')
soc.send('#Recovered Transmission\n')
dataSend = True
except:
print("unable to connect")
dataSend = False
if(value == 0):
msg, CVframe, maskFrame = vision(cap)
try:
soc.send(msg)
print(msg)
except:
print 'data send failed'
dataSend = False
frame = cv2.imencode('.jpg',CVframe,[IMWRITE_JPEG_QUALITY, 10])[1].tostring()
elif(value == 1):
ret, CVframe = cap_two.read()
# if(ret == False):
# frame = polarBearError
# else:
frame = cv2.imencode('.jpg',CVframe,[IMWRITE_JPEG_QUALITY, 10])[1].tostring()
elif(value ==2):
ret, CVframe = cap_three.read()
# if(ret == False):
# frame = pandaBearError
# else:
frame = cv2.imencode('.jpg',CVframe,[IMWRITE_JPEG_QUALITY, 10])[1].tostring()
else:
msg, CVframe, maskFrame = vision(cap)
try:
soc.send(msg)
except:
print 'data send failed'
dataSend = False
frame = cv2.imencode('.jpg',maskFrame,[IMWRITE_JPEG_QUALITY, 10])[1].tostring()
yield (b'--frame\r\n'
b'Content-Type: image/jpeg\r\n\r\n' + frame + b'\r\n')
if cv2.waitKey(1) & 0xFF == ord('q'):
break
def gen(value):
"""Video streaming generator function."""
soc = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
soc.settimeout(2)
dataSend = True
# connect to remote host
try:
soc.connect((host, port))
print "found target"
soc.send("#Found you\n")
except:
print ("unable to connect")
while 1:
if dataSend == False:
try:
soc.connect((host, port))
print ("found target")
soc.send("#Recovered Transmission\n")
dataSend = True
except:
print ("unable to connect")
if value == 0:
msg, CVframe, maskFrame = vision(cap)
try:
soc.send(msg)
except:
print "data send failed"
dataSend = False
frame = cv2.imencode(".jpg", CVframe, [IMWRITE_JPEG_QUALITY, 10])[1].tostring()
elif value == 1:
ret, CVframe = cap_two.read()
# if(ret == False):
# frame = polarBearError
# else:
frame = cv2.imencode(".jpg", CVframe, [IMWRITE_JPEG_QUALITY, 10])[1].tostring()
elif value == 2:
ret, CVframe = cap_three.read()
# if(ret == False):
# frame = pandaBearError
# else:
frame = cv2.imencode(".jpg", CVframe, [IMWRITE_JPEG_QUALITY, 10])[1].tostring()
else:
msg, CVframe, maskFrame = vision(cap)
# try:
# soc.send(msg)
# except:
# print 'Lost Connection with Roborio'
# soc.connect((host, port))
# print('found target')
# soc.send("#Found you\n")
frame = cv2.imencode(".jpg", maskFrame, [IMWRITE_JPEG_QUALITY, 10])[1].tostring()
yield (b"--frame\r\n" b"Content-Type: image/jpeg\r\n\r\n" + frame + b"\r\n")
if cv2.waitKey(1) & 0xFF == ord("q"):
break
def svhn(pil_image, province):
img = cv2.cvtColor(np.array(pil_image), cv2.COLOR_RGB2BGR)
if province == 'nacao':
img = nacao(img)
imgstr = ''
if pil_image.format == 'JPEG':
imgstr = cv2.imencode('.jpg', img)[1].tostring()
elif pil_image.format == 'PNG':
imgstr = cv2.imencode('.png', img)[1].tostring()
return base64.b64encode(imgstr)
def encodeImage(self, img):
""" Encodes the np-array given by img into jpeg-data that can be sent over http """
if isinstance(img, type(None)) or img.size == 0:
_, data = cv2.imencode('.jpg', np.zeros((1), np.uint8))
else:
with self.dataLock:
_, data = cv2.imencode('.jpg', img)
jpeg_data = data.tostring()
return jpeg_data
def callback_right_with_3D(self, ros_data):
'''Callback function of subscribed topic.
Here images get converted and features detected'''
self.right_header = ros_data.header.stamp.secs
#### direct conversion to CV2 ####
np_arr = np.fromstring(ros_data.data, np.uint8)
image_np = cv2.imdecode(np_arr, 1)
# convert np image to grayscale
gray_image = cv2.cvtColor(image_np, cv2.COLOR_BGR2GRAY)
if self.right_header == self.left_header and self.group_4_recieved:
s_t = time.clock()
print self.right_header, self.left_header
self.group_4_recieved = False
left_gray = cv2.cvtColor(self.left_image, cv2.COLOR_BGR2GRAY)
group_4_out = self.group_4
# PointCloudPose()
# group_4_out.header = self.group_4.header
#std_msgs/Int16 pose_id
#std_msgs/Int16 pose_id_max
#sensor_msgs/CompressedImage image_left
#sensor_msgs/CompressedImage image_right
#geometry_msgs/Pose spin_center_pose
#sensor_msgs/PointCloud2 carmine_pointcloud
#geometry_msgs/Pose carmine_pose
#sensor_msgs/PointCloud2 bumblebee_pointcloud
#geometry_msgs/Pose bumblebee_pose_left
#geometry_msgs/Pose bumblebee_pose_right
msg = CompressedImage()
msg.header.stamp = rospy.Time.now()
msg.format = "jpeg"
msg.data = np.array(cv2.imencode('.jpg', left_gray)[1]).tostring()
# Publish new image
group_4_out.image_left = msg
msg = CompressedImage()
msg.header.stamp = rospy.Time.now()
msg.format = "jpeg"
msg.data = np.array(cv2.imencode('.jpg', gray_image)[1]).tostring()
# Publish new image
group_4_out.image_right = msg
#group_4_out.bumblebee_pointcloud = pcloud
self.pub_group.publish(group_4_out)
print "save time", time.clock() - s_t
def convert(image, input_type='cv2', output_type='ros'):
assert input_type in ['cv2', 'ros', 'mongo'] and output_type in ['cv2', 'ros', 'mongo']
if input_type == 'mongo':
image = cv2.imdecode(np.array(image, dtype=np.uint8), 1)
if input_type == 'ros':
image = cv2.imdecode(np.fromstring(image, np.uint8), cv2.CV_LOAD_IMAGE_COLOR)
if output_type == 'ros':
return np.array(cv2.imencode('.jpg', image)[1]).tostring()
if output_type == 'mongo':
return np.array(cv2.imencode('.jpg', image)[1]).tolist()
if output_type == 'cv2':
return image
def set_image(self, data_bytes):
data = np.frombuffer(data_bytes, dtype="uint8")
cv2im = cv2.imdecode(data, 1)
(cv2green, cv2red) = split(cv2im)
_, green = cv2.imencode(".jpeg", cv2green)
_, red = cv2.imencode(".jpeg", cv2red)
_, st_img = cv2.imencode(".jpeg", stereo(cv2green, cv2red))
self.original_holder.set_image(data)
self.green_holder.set_image(green)
self.red_holder.set_image(red)
self.stereo_holder.set_image(st_img)
def get_image(request):
url = request.GET.get('url', None)
if url:
# Try to get existing source first
try:
src = Source.objects.get(url=url)
except Exception:
src = None
# No existing source or not active? Try to open and get one frame
if not src or src.active == False:
try:
if url.isdigit():
video_capture = cv2.VideoCapture(int(url))
else:
video_capture = cv2.VideoCapture(url)
ret, frame = video_capture.read()
if frame.shape[1]!=640 or frame.shape[0]!=480:
frame = cv2.resize(frame, (640,480))
r,buf=cv2.imencode(".jpg",frame)
JpegData=buf.tostring()
video_capture.release()
except Exception:
video_capture.release()
return HttpResponse("")
return HttpResponse(JpegData, content_type="image/png")
# Source exists and is active? Grab frame from mmapped file
else:
try:
sink = src.raw_sink
filename = os.path.join(settings.PROJECT_ROOT,"run","sinks",sink.short_id)
if not os.path.isfile(filename):
return HttpResponse("")
size = os.path.getsize(filename)
f = open(filename, 'r')
m = mmap.mmap(f.fileno(), 0, access=mmap.ACCESS_READ)
capture = m
capture.seek(0, os.SEEK_SET)
dt = np.dtype((np.uint8, 3))
img = np.fromstring(capture.read(size), dt)
img = np.reshape(img,(480,640,3))
r,buf=cv2.imencode(".jpg",img)
JpegData=buf.tostring()
f.close()
except Exception:
return HttpResponse("")
return HttpResponse(JpegData, content_type="image/png")
else:
return HttpResponse("")
def type4_cut(pil_image, province):
# cut the image and encode to base64 strings
img = cv2.cvtColor(np.array(pil_image), cv2.COLOR_RGB2BGR)
mapper = {'chongqing':chq, 'gansu':gs, 'jiangxi':jx, 'ningxia':nx, 'tianjin':tj,
'shan3xi':small, 'sichuan':small, 'xinjiang':small, 'beijing':beijing, 'hubei':hubei}
patches = mapper[province](img)
for i in range(len(patches)):
imgstr = ''
if pil_image.format == 'JPEG':
imgstr = cv2.imencode('.jpg', patches[i])[1].tostring()
elif pil_image.format == 'PNG':
imgstr = cv2.imencode('.png', patches[i])[1].tostring()
patches[i] = base64.b64encode(imgstr)
return patches
def read(self, extension=None, quality=None):
if quality is None:
quality = self.context.config.QUALITY
options = None
extension = extension or self.extension
try:
if FORMATS[extension] == 'JPEG':
options = [cv2.IMWRITE_JPEG_QUALITY, quality]
except KeyError:
# default is JPEG so
options = [cv2.IMWRITE_JPEG_QUALITY, quality]
try:
if FORMATS[extension] == 'WEBP':
options = [cv2.IMWRITE_WEBP_QUALITY, quality]
except KeyError:
options = [cv2.IMWRITE_JPEG_QUALITY, quality]
success, buf = cv2.imencode(extension, self.image, options or [])
data = buf.tostring()
if FORMATS[extension] == 'JPEG' and self.context.config.PRESERVE_EXIF_INFO:
if hasattr(self, 'exif') and self.exif != None:
img = JpegFile.fromString(data)
img._segments.insert(0, ExifSegment(self.exif_marker, None, self.exif, 'rw'))
data = img.writeString()
return data
def pack_img(header, img, quality=95, img_fmt='.jpg'):
"""Pack an image into ``MXImageRecord``.
Parameters
----------
header : IRHeader
Header of the image record.
``header.label`` can be a number or an array.
img : numpy.ndarray
image to pack
quality : int
Quality for JPEG encoding in range 1-100, or compression for PNG encoding in range 1-9.
img_fmt : str
Encoding of the image (.jpg for JPEG, .png for PNG).
Returns
-------
s : str
The packed string.
"""
assert cv2 is not None
jpg_formats = ['.JPG', '.JPEG']
png_formats = ['.PNG']
encode_params = None
if img_fmt.upper() in jpg_formats:
encode_params = [cv2.IMWRITE_JPEG_QUALITY, quality]
elif img_fmt.upper() in png_formats:
encode_params = [cv2.IMWRITE_PNG_COMPRESSION, quality]
ret, buf = cv2.imencode(img_fmt, img, encode_params)
assert ret, 'failed to encode image'
return pack(header, buf.tostring())
def do_GET(self):
print self.path
if self.path.endswith('.mjpg'):
self.send_response(200)
self.send_header('Content-type','multipart/x-mixed-replace; boundary=--jpgboundary')
self.end_headers()
while True:
try:
rc,img = capture.read()
if not rc:
continue
imgRGB = cv2.cvtColor(img,cv2.COLOR_BGR2RGB)
r, buf = cv2.imencode(".jpg",imgRGB)
self.wfile.write("--jpgboundary\r\n")
self.send_header('Content-type','image/jpeg')
self.send_header('Content-length',str(len(buf)))
self.end_headers()
self.wfile.write(bytearray(buf))
self.wfile.write('\r\n')
time.sleep(0.5)
except KeyboardInterrupt:
break
return
if self.path.endswith('.html') or self.path=="/":
self.send_response(200)
self.send_header('Content-type','text/html')
self.end_headers()
self.wfile.write('<html><head></head><body>')
self.wfile.write('<img src="http://127.0.0.1:9090/cam.mjpg"/>')
self.wfile.write('</body></html>')
return
def get_frame(self):
success, image = self.video.read()
# We are using Motion JPEG, but OpenCV defaults to capture raw images,
# so we must encode it into JPEG in order to correctly display the
# video stream.
ret, jpeg = cv2.imencode('.jpg', image)
return jpeg.tobytes()
def test_imencode(self):
a = np.zeros((480, 640), dtype=np.uint8)
flag, ajpg = cv2.imencode("img_q90.jpg", a, [cv2.IMWRITE_JPEG_QUALITY, 90])
self.assertEqual(flag, True)
self.assertEqual(ajpg.dtype, np.uint8)
self.assertGreater(ajpg.shape[0], 1)
self.assertEqual(ajpg.shape[1], 1)
def process(self, stream):
data = cv2.imdecode(np.fromstring(stream.getvalue(), dtype=np.uint8), 1)
(orig_h, orig_w) = data.shape[:2]
new_h = orig_h / 2
new_w = orig_w / 2
new_dims = (new_w, new_h)
# draw a test rectangle
cv2.rectangle(data, (10, 400), (300, 300), (0, 255, 0), 1)
# draw a test square
data[10:20, 10:20] = (255, 0, 0)
(B, G, R) = cv2.split(data)
zeros = np.zeros(data.shape[:2], dtype = "uint8")
Brz = self.prepare_channel(B, zeros, zeros, new_dims)
Grz = self.prepare_channel(zeros, G, zeros, new_dims)
Rrz = self.prepare_channel(zeros, zeros, R, new_dims)
Orz = self.prepare_channel(B, G, R, new_dims)
row1 = np.concatenate((Orz, Rrz), axis = 1)
row2 = np.concatenate((Brz, Grz), axis = 1)
data = np.concatenate((row1, row2), axis = 0)
return cv2.imencode('.jpg', data)[1].tostring()
def do_GET(self):
global finalImg
print self.path
if self.path.endswith('.mjpg'):
self.send_response(200)
self.send_header('Content-type','multipart/x-mixed-replace; boundary=--jpgboundary')
self.end_headers()
while True:
try:
img = classify_image(finalImg)
r, buf = cv2.imencode(".jpg",img)
self.wfile.write("--jpgboundary\r\n")
self.send_header('Content-type','image/jpeg')
self.send_header('Content-length',str(len(buf)))
self.end_headers()
self.wfile.write(bytearray(buf))
self.wfile.write('\r\n')
time.sleep(0.05)
except KeyboardInterrupt:
break
except:
pass
return
if self.path.endswith('.html') or self.path=="/":
self.send_response(200)
self.send_header('Content-type','text/html')
self.end_headers()
self.wfile.write('<html><head></head><body>')
self.wfile.write('<img src="./cam.mjpg" />')
self.wfile.write('</body></html>')
return
def post_media(self, img):
result, img_png = cv2.imencode('.png', img)
if not result:
IkaUtils.dprint('%s: Failed to encode the image (%s)' %
(self, img.shape))
return None
files = { "media": img_png.tostring() }
CK = self._preset_ck if self.consumer_key_type == 'ikalog' else self.consumer_key
CS = self._preset_cs if self.consumer_key_type == 'ikalog' else self.consumer_secret
from requests_oauthlib import OAuth1Session
twitter = OAuth1Session(
CK, CS, self.access_token, self.access_token_secret
)
req = twitter.post(
self.url_media,
files=files,
verify=self._get_cert_path()
)
if req.status_code == 200:
return json.loads(req.text)['media_id']
IkaUtis.dprint('%s: Failed to post media.' % self)
return None
def openCV_to_kivy(cvImage):
# encode the OpenCV image in PNG format
_, imPNG = cv.imencode(".png", cvImage)
# create a binary data stream for reading that data
data = io.BytesIO(imPNG.tobytes())
# create a Kivy Core Image data structure to hold that data
return CoreImage(data, ext="png")
def thumbnail(self, filename):
cap = cv2.VideoCapture(self.media_path + "/" + filename)
ret, frame = cap.read()
ret, jpg = cv2.imencode(".jpg", frame)
cap.release()
cherrypy.response.headers['Content-Type'] = 'image/jpeg'
return jpg.tostring()
def hash_encode_image(image, extension):
""" Encode the image, get the hash and return the hash with
encoded image """
img = cv2.imencode(extension, image)[1] # pylint: disable=no-member
f_hash = sha1(
cv2.imdecode(img, cv2.IMREAD_UNCHANGED)).hexdigest() # pylint: disable=no-member
return f_hash, img
def captureFrame():
print("Capturing 1")
cap = cv2.VideoCapture(-1)
if ( not cap.isOpened()):
print("Camera Failed to open")
return b'0'
cap.set(3, 640)
cap.set(4, 480)
trials = 3
frame = np.array([0])
stat = False
while ( not stat and trials > 0 and (frame == 0).all() ):
time.sleep(.3) # Sleep for 200 ms for camera to turn on;
stat, frame = cap.read()
trials -= 1;
cap.release()
ret = b'0'
if (stat):
ret = cv2.imencode('.jpg', frame)[1].tobytes()
return ret;
def process_image(image_buffer):
'''
receive image from client. remove lines. send processed image back
'''
# print to debug
print('type:', type(image_buffer), 'len:', len(image_buffer))
# construct image from binary buffer/file stream or whatever abstraction you see
np_array = np.frombuffer(image_buffer, dtype='uint8')
print('type:', type(np_array), 'len:', len(np_array))
# read image from raw data
grayscale_image = cv2.imdecode(np_array, cv2.IMREAD_GRAYSCALE)
# process the received image
threshed = process_an_image(grayscale_image)
remove_lines(threshed)
# cv2.imshow('', threshed) # debug
# cv2.waitKey()
ret, buffer_of_image = cv2.imencode('.jpg', threshed)
buffer_string = buffer_of_image.tostring()
list_of_byte = list(buffer_string)
print('type:', type(buffer_of_image), 'len:', len(buffer_of_image)) # debug
print('type:', type(list_of_byte), 'len:', len(list_of_byte))
print('type:', type(buffer_string), 'len:', len(buffer_string))
return buffer_string
def stream_image(fn):
"""Video streaming generator function."""
while True:
frame = ""
try:
img = cv2.imread(fn,cv2.IMREAD_COLOR)
b,frame = cv2.imencode(".jpg", img)
frame = frame.tostring()
# with open(fn, 'rb') as f:
# frame = f.read()
except Exception as e:
print(str(e))
img = Image.new("RGB", (512, 256))
draw = ImageDraw.Draw(img)
pos = float(time.time() * 1000) / 1000
pos = (1 + math.sin(pos)) / 2
pos = pos * (img.size[0] - 50)
draw.multiline_text((pos, img.size[1] / 2),
"NO IMAGE",
fill="white")
del draw
b = io.BytesIO()
img.save(b, "JPEG")
b.seek(0)
frame = b.read()
yield (b'--frame\r\n'
b'Content-Type: image/jpeg\r\n\r\n' + frame + b'\r\n')
# Set up the necessary objects for detection
detector = setup_detector()
table_factory = TableSetup(cfg)
# Run the detection algorithm
detections = detector.detect()
i = 0
for image, detection in detections:
# Convert the detection to a SnookerTable object if possible
table = table_factory.create_table(detection)
if table is None:
continue
print "Frame {0}".format(i)
# Get the json and log details to send to the RabbitMQ server
json = table.to_json()
messenger.send(json, "json")
messenger.send(json, "log")
print "Snooker ball count: {0}".format(len(table.balls))
# Convert the frame to send to the server
img = cv2.imencode('.jpg', image)[1].tostring()
messenger.send(img, "stream")
print "Image byte size: {0}".format(len(img))
i += 1
print "Stream complete"
messenger.disconnect()
def encode_toString(img):
retval, buffer = cv2.imencode('.jpg', img)
data = base64.b64encode(buffer)
return data
def run(self):
while True:
camera, frame_time, tracked_objects = self.tracked_objects_queue.get(
)
config = self.config[camera]
best_objects = self.camera_data[camera]['best_objects']
current_object_status = self.camera_data[camera]['object_status']
self.camera_data[camera]['tracked_objects'] = tracked_objects
self.camera_data[camera]['current_frame_time'] = frame_time
###
# Draw tracked objects on the frame
###
current_frame = self.plasma_client.get(f"{camera}{frame_time}")
if not current_frame is plasma.ObjectNotAvailable:
# draw the bounding boxes on the frame
for obj in tracked_objects.values():
thickness = 2
color = COLOR_MAP[obj['label']]
if obj['frame_time'] != frame_time:
thickness = 1
color = (255, 0, 0)
# draw the bounding boxes on the frame
box = obj['box']
draw_box_with_label(
current_frame,
box[0],
box[1],
box[2],
box[3],
obj['label'],
f"{int(obj['score']*100)}% {int(obj['area'])}",
thickness=thickness,
color=color)
# draw the regions on the frame
region = obj['region']
cv2.rectangle(current_frame, (region[0], region[1]),
(region[2], region[3]), (0, 255, 0), 1)
if config['snapshots']['show_timestamp']:
time_to_show = datetime.datetime.fromtimestamp(
frame_time).strftime("%m/%d/%Y %H:%M:%S")
cv2.putText(current_frame,
time_to_show, (10, 30),
cv2.FONT_HERSHEY_SIMPLEX,
fontScale=.8,
color=(255, 255, 255),
thickness=2)
###
# Set the current frame
###
self.camera_data[camera]['current_frame'] = current_frame
# delete the previous frame from the plasma store and update the object id
if not self.camera_data[camera]['object_id'] is None:
self.plasma_client.delete(
self.camera_data[camera]['object_id'])
self.camera_data[camera]['object_id'] = f"{camera}{frame_time}"
###
# Maintain the highest scoring recent object and frame for each label
###
for obj in tracked_objects.values():
# if the object wasn't seen on the current frame, skip it
if obj['frame_time'] != frame_time:
continue
if obj['label'] in best_objects:
now = datetime.datetime.now().timestamp()
# if the object is a higher score than the current best score
# or the current object is more than 1 minute old, use the new object
if obj['score'] > best_objects[obj['label']]['score'] or (
now -
best_objects[obj['label']]['frame_time']) > 60:
obj['frame'] = np.copy(
self.camera_data[camera]['current_frame'])
best_objects[obj['label']] = obj
else:
obj['frame'] = np.copy(
self.camera_data[camera]['current_frame'])
best_objects[obj['label']] = obj
###
# Report over MQTT
###
# count objects with more than 2 entries in history by type
obj_counter = Counter()
for obj in tracked_objects.values():
if len(obj['history']) > 1:
obj_counter[obj['label']] += 1
# report on detected objects
for obj_name, count in obj_counter.items():
new_status = 'ON' if count > 0 else 'OFF'
if new_status != current_object_status[obj_name]:
current_object_status[obj_name] = new_status
self.client.publish(
f"{self.topic_prefix}/{camera}/{obj_name}",
new_status,
retain=False)
# send the best snapshot over mqtt
best_frame = cv2.cvtColor(best_objects[obj_name]['frame'],
cv2.COLOR_RGB2BGR)
ret, jpg = cv2.imencode('.jpg', best_frame)
if ret:
jpg_bytes = jpg.tobytes()
self.client.publish(
f"{self.topic_prefix}/{camera}/{obj_name}/snapshot",
jpg_bytes,
retain=True)
self.client.publish(
f"{self.topic_prefix}/{camera}/{obj_name}/region",
"{},{},{},{}".format(region[0], region[1],
region[2], region[3]),
retain=False)
self.client.publish(
f"{self.topic_prefix}/{camera}/{obj_name}/box",
"{},{},{},{}".format(box[0], box[1], box[2],
box[3]),
retain=False)
# expire any objects that are ON and no longer detected
expired_objects = [
obj_name for obj_name, status in current_object_status.items()
if status == 'ON' and not obj_name in obj_counter
]
for obj_name in expired_objects:
current_object_status[obj_name] = 'OFF'
self.client.publish(f"{self.topic_prefix}/{camera}/{obj_name}",
'OFF',
retain=False)
# send updated snapshot over mqtt
best_frame = cv2.cvtColor(best_objects[obj_name]['frame'],
cv2.COLOR_RGB2BGR)
ret, jpg = cv2.imencode('.jpg', best_frame)
if ret:
jpg_bytes = jpg.tobytes()
self.client.publish(
f"{self.topic_prefix}/{camera}/{obj_name}/snapshot",
jpg_bytes,
retain=True)
self.client.publish(
f"{self.topic_prefix}/{camera}/{obj_name}/region",
"{},{},{},{}".format(region[0], region[1], region[2],
region[3]),
retain=False)
self.client.publish(
f"{self.topic_prefix}/{camera}/{obj_name}/box",
"{},{},{},{}".format(box[0], box[1], box[2], box[3]),
retain=False)
def classify_image (event, context):
try:
content_type_header = event['headers']['content-type']
body = base64.b64decode(event["body"])
picture = decoder.MultipartDecoder(body, content_type_header).parts[0]
im_arr = np.frombuffer(picture.content, dtype=np.uint8)
im = cv2.imdecode(im_arr, flags=cv2.IMREAD_COLOR)
points = fbc.getLandmarks(faceDetector, landmarkDetector, im)
points = np.array(points)
im = np.float32(im)/255.0
h = 600
w = 600
imNorm, points = fbc.normalizeImagesAndLandmarks((h, w), im, points)
imNorm = np.uint8(imNorm*255)
filename = picture.headers[b'Content-Disposition'].decode().split(';')[1].split('=')[1]
if len(filename) < 4:
filename = picture.headers[b'Content-Disposition'].decode().split(';')[2].split('=')[1]
print ('all done')
return {
"statusCode": 200,
"headers": {
'Content-Type': 'application/json',
'Access-Control-Allow-Origin': '*',
"Access-Control-Allow-Credentials": True
},
"body": json.dumps({'file': filename.replace('"', ''), 'aligned_image': str (base64.b64encode (cv2.imencode ('.jpg', imNorm)[1])) })
}
except Exception as e:
print(repr(e))
return {
"statusCode": 500,
"headers": {
'Content-Type': 'application/json',
'Access-Control-Allow-Origin': '*',
"Access-Control-Allow-Credentials": True
},
"body": json.dumps({"error": repr(e)})
}
def save_inp_image(self, img, group):
img_str = cv2.imencode('.png', img)[1]
self.save('input', img_str, group)
def frame_tobytes(self):
''' Encode the output frame into bytes.
'''
return cv2.imencode('.jpg', self.visualize())[1].tobytes()
## setup logging
import logging
logging.basicConfig(level = logging.INFO)
## import the package
import numpy as np
import cv2
import sys
cap = cv2.VideoCapture(0)
while(True):
# Capture frame-by-frame
ret, frame = cap.read()
# Our operations on the frame come here
gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
# Display the resulting frame
success,image = cap.read()
print(cv2.imencode('.jpg', image)[1].tobytes())
# When everything done, release the capture
cap.release()
cv2.destroyAllWindows()
if writer is None:
# initialize our video writer
fourcc = cv2.VideoWriter_fourcc(*"MJPG")
writer = cv2.VideoWriter(args["output"], fourcc, 30,
(frame.shape[1], frame.shape[0]), True)
# some information on processing single frame
if total > 0:
elap = (end - start)
print("[INFO] single frame took {:.4f} seconds".format(elap))
print("[INFO] estimated total time to finish: {:.4f}".format(
elap * total))
#write the output frame to disk
writer.write(frame)
imgbytes = cv2.imencode('.png', frame)[1].tobytes() # ditto
if not win_started:
win_started = True
layout = [[
sg.Text('Yolo Playback in PySimpleGUI Window', size=(30, 1))
], [sg.Image(data=imgbytes, key='_IMAGE_')],
[
sg.Text('Confidence'),
sg.Slider(range=(0, 1),
orientation='h',
resolution=.1,
default_value=.5,
size=(15, 15),
key='confidence'),
sg.Text('Threshold'),
def mark_diff(self, diff_image):
"""
Save 'difference' pixels to a picture
"""
is_success, buffer = cv2.imencode(".png", diff_image)
return io.BytesIO(buffer).getvalue()
if len(sys.argv) > 1:
broker_uri = sys.argv[1]
topic_id = sys.argv[2]
channel = Channel(broker_uri)
exporter = ZipkinExporter(
service_name='CameraGateway.{}'.format(topic_id),
host_name='localhost',
port=9411,
transport=BackgroundThreadTransport(max_batch_size=100),
)
image = cv2.imread('../image.png')
for k in range(10):
tracer = Tracer(exporter)
with tracer.span(name='image') as span:
cimage = cv2.imencode(ext='.jpeg',
img=image,
params=[cv2.IMWRITE_JPEG_QUALITY, 80])
data = cimage[1].tobytes()
im = Image(data=data)
msg = Message()
msg.topic = 'CameraGateway.{}.Frame'.format(topic_id)
msg.inject_tracing(span)
msg.metadata.update({'image_id': k})
msg.pack(im)
channel.publish(msg)
log.info('Message {} published', k)
time.sleep(0.250)
def root():
# removing all previous files in folder before start processing
output_folder = 'static/img/temp/'
for file in glob.glob(output_folder + '*'):
os.remove(file)
# on POST handle upload
if request.method == 'POST':
# get file details
file_data = request.files.get('file')
if file_data is None:
err_msg = 'No input image was provided.'
logging.error(err_msg)
return render_template('index.html', error_msg=err_msg)
# read image from string data
file_request = file_data.read()
# convert string data to numpy array
np_inp_image = np.fromstring(file_request, np.uint8)
img = cv2.imdecode(np_inp_image, cv2.IMREAD_UNCHANGED)
_, image_encoded = cv2.imencode('.jpeg', img)
# TODO R1: review inference request payload
# Required inference request parameter: image (JPG/PNG encoded)
files = {
'file': image_encoded.tostring(),
'Content-Type': 'multipart/form-data',
}
# TODO T1: replace model URL placeholder
# Add model endpoint
model_url = args.ml_endpoint.rstrip('/') + '**TODO**'
# Send image file form to model endpoint for prediction
try:
results = requests.post(url=model_url, files=files)
except Exception as e:
err_msg_temp = 'Prediction request to {} failed: {}'
err_msg = err_msg_temp.format(model_url, 'Check log for details.')
logging.error(err_msg_temp.format(model_url, str(e)))
return render_template("index.html", error_msg=err_msg)
# surface any prediction errors to user
if results.status_code != 200:
err_msg = ('Prediction request returned status code {} ' +
'and message {}').format(results.status_code,
results.text)
logging.error(err_msg)
return render_template('index.html', error_msg=err_msg)
# extract prediction from json return
output_data = results.json()
# log output in debug
logging.debug('\n' + pformat(output_data))
# TODO T2: replace placeholder with appropriate JSON key
# Extraction prediction result
result = output_data['**TODO**']
if len(result) == 0:
msg = 'No objects detected, try uploading a new image'
return render_template('index.html', error_msg=msg)
else:
# save the output image to return
file_name = (str(randint(0, 999999)) + '.jpg')
output_name = output_folder + '/' + file_name
im = Image.fromarray(img)
im = im.convert("L")
newsize = (300, 300)
im = im.resize(newsize)
im.save(output_name)
return render_template('index.html',
image_name=output_name,
result_1=result[0]['prediction'])
else:
# on GET return index.html
return render_template('index.html')
res, score = '', 0.0
i = 0
mem = ''
consecutive = 0
textInput = ''
while True:
ret, img = cap.read()
img = cv2.flip(img, 1)
if ret:
x1, y1, x2, y2 = 200, 200, 400, 400
img_cropped = img[y1:y2, x1:x2]
c += 1
image_data = cv2.imencode('.jpg', img_cropped)[1].tostring()
a = cv2.waitKey(1) # waits to see if `esc` is pressed
if i == 4:
res_tmp, score = predict(image_data)
res = res_tmp
i = 0
if mem == res:
consecutive += 1
else:
consecutive = 0
if consecutive == 2 and res not in ['nothing']:
if res == 'space':
textInput += ' '
elif res == 'del':
def img2str(img):
return cv2.imencode('.jpg', img)[1].tostring()
if img.shape[2] == 3:
img = img[:, :, [2, 1, 0]]
elif img.shape[2] == 4:
img = img[:, :, [2, 1, 0, 3]]
img = torch.from_numpy(np.transpose(img, (2, 0, 1))).float()
img_LR = img.unsqueeze(0)
img_LR = img_LR.to(device)
output = model(img_LR).data.squeeze(0).float().cpu().clamp_(0, 1).numpy()
if output.shape[0] == 3:
output = output[[2, 1, 0], :, :]
elif output.shape[0] == 4:
output = output[[2, 1, 0, 3], :, :]
output = np.transpose(output, (1, 2, 0))
output = (output * 255.0).round()
if passAsString == True:
buffer = cv2.imencode(".png", output)[1]
data = base64.b64encode(buffer)
print(data)
continue
newpath = base
printpath = ''
if mode == '1' or mode == '2':
baseinput = os.path.splitext(os.path.basename(name))[0]
baseinput = re.search('(.*)(_tile-[0-9]+)', baseinput, re.IGNORECASE).group(1)
modelname = os.path.splitext(os.path.basename(model_path))[0]
if mode == '1':
os.makedirs('{1:s}/Images/{0:s}/'.format(baseinput, output_folder), exist_ok=True)
newpath = '{3:s}/Images/{0:s}/[{2:s}]_{1:s}.png'.format(baseinput, base, modelname, output_folder)
if mode == '2':
os.makedirs('{1:s}/Models/{0:s}/'.format(modelname, output_folder), exist_ok=True)
def resize_512():
cv2.namedWindow("test", 0)
cv2.resizeWindow("test", 512, 512)
num_dic = {'a': 0, 'b': 1, 'c': 2, 'd': 3, 'e': 4, 'f': 5}
org_path = r'F:\1_sheng\答题卡拍照'
ner_path = r'F:\1_sheng\card_512'
all_index = 0
txt_handle = open('./train.txt', 'w')
for folder in range(1, 10, 1):
second_path = os.path.join(org_path, str(folder))
all_json_list = glob.glob(os.path.join(second_path, "*.json"))
for i, one_json_path in enumerate(all_json_list):
with codecs.open(one_json_path,
'r',
encoding='utf-8',
errors='ignore') as f:
jsondict = json.load(f)
imagePath = one_json_path.replace(".json", ".jpg")
if not os.path.exists(imagePath):
imagePath = one_json_path.replace(".json", ".png")
print(imagePath, "no exit, png")
elif not os.path.exists(imagePath):
imagePath = one_json_path.replace(".json", ".jpeg")
print(imagePath, "no exit, jpeg")
elif not os.path.exists(imagePath):
print(imagePath, "no exit, 注意后缀名---------------")
break
pic = jsondict['imageData']
pic = base64.b64decode(pic)
# print(pic)
pic = np.fromstring(pic, np.uint8)
img = cv2.imdecode(pic, cv2.COLOR_BGR2RGB)
# img = cv_imread(imagePath)
height = img.shape[0]
width = img.shape[1]
resize_img = cv2.resize(img, (512, 512))
gt_content = ""
for one_point in jsondict['shapes']:
label = one_point['label']
# print(label, "one_point['points']: ", one_point['points'])
x = int(one_point['points'][0][0])
y = int(one_point['points'][0][1])
label_index = num_dic[label]
# print(x/width, y/height)
x = int((x / width) * 512)
y = int((y / height) * 512)
# resize_img = cv2.circle(resize_img, (x, y), 3, [0, 255, 0], 5)
gt_content = gt_content + ' {},{},{}'.format(
str(label_index), str(x), str(y))
img_write_path = os.path.join(ner_path, str(folder),
os.path.split(imagePath)[1])
cv2.imencode('.jpg', resize_img)[1].tofile(img_write_path)
write_content = img_write_path + gt_content
txt_handle.write(write_content + '\n')
print(all_index, write_content)
# cv2.imshow("test", resize_img)
# cv2.waitKey(0)
all_index += 1
txt_handle.close()
def get_frame(self):
(self.grabbed, self.frame) = self.video.read()
image = self.frame
ret, jpeg = cv2.imencode('.jpg', image)
return jpeg.tobytes()
linger_ms=5,
max_request_size=2097152,
compression_type='gzip'
)
while(True):
### read a frame from camera
ret, frame = cap.read()
frame = cv2.flip(frame, 1)
### prepare frame attributes
timestamp = datetime.datetime.now()
rows, cols, channels = frame.shape
frame = cv2.resize(frame, (480, int(480.0*rows/cols)), interpolation = cv2.INTER_LINEAR)
img_encode = cv2.imencode('.png', frame)[1]
data_encode = np.array(img_encode)
# cv2.imshow('frame', frame)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
### construc JSON message
msg_dict = {
"cameraId":"vid-01",
"timestamp":timestamp.strftime('%Y-%m-%d %H:%M:%S'),
"data":base64.b64encode(data_encode).decode('utf-8')
}
msg_json = json.dumps(msg_dict)
# print(msg_json)
face_client = FaceClient(ENDPOINT, CognitiveServicesCredentials(KEY))
# Initialize camera by CV2
cam = cv2.VideoCapture(0)
cv2.namedWindow("face")
img_counter = 0
# While loop to continuously process the video frames
while True:
ret, frame = cam.read()
if not ret:
print("failed to grab frame")
break
cv2.imshow("face", frame)
# Post video frames to Azure Face Service to obtain face landmarks
image = cv2.imencode('.jpg', frame)[1].tostring()
subscription_key = KEY
face_api_url = "https://southeastasia.api.cognitive.microsoft.com/face/v1.0/detect"
headers = {'Content-Type': 'application/octet-stream', 'Ocp-Apim-Subscription-Key': subscription_key}
params = {'returnFaceId': 'true', 'returnFaceLandmarks': 'true'}
response = requests.post(face_api_url, params=params, headers=headers, data=image)
response.raise_for_status()
faces = response.json()
print(faces)
# Parse collected face landmarks into variables
for face in faces:
flm = face['faceLandmarks']
pupilLeft = flm['pupilLeft']
pupilRight = flm['pupilRight']
noseTip = flm['noseTip']
def getImage(self):
total = T.total #total number of bounding boxes
boundingBoxes = T.boundingBoxes #stored bounding box coordinates from last frame
prevImage = T.prevImage #stored image from previous frame
file = self.stream.read()
encoded_string = base64.b64encode(file)
arr = np.asarray(bytearray(file), dtype=np.uint8)
img = cv2.imdecode(arr, -1)
#filter out background
img2 = img.copy()
img2 = self.filter2(79, 104, 43, 240, 141, 108, 351, 195, img2) #hacked solution to black out the non-essential parts of the image
sensitivity = 0.4 #threshold to filter out detections
boxes, scores, classes, num = C.getClassification(img2) #runs the image through ssd-mobilenet
limit = 0
for i in range(scores[0].shape[0]):
limit = i
if scores[0][i] < sensitivity:
break
nBoxes = boxes[0][0:limit]
nScores = scores[0][0:limit]
nClasses = classes[0][0:limit]
currentBoxes = []
for box1 in nBoxes:
x1 = min(351,int(round(box1[1]*352)))
y1 = min(239,int(round(box1[0]*240)))
x2 = min(351,int(round(box1[3]*352)))
y2 = min(239,int(round(box1[2]*240)))
print((x1, x2, y1, y2))
currentBoxes.append((x1, x2, y1, y2))
#currentBoxes now holds the coordinates of the bounding boxes for this frame
#TODO: Run the current bounding boxes through VGG
#EXAMPLE
for coord in currentBoxes:
(x1, x2, y1, y2) = coord
boundingBox = img[y1:y2, x1:x2, :]
#Determine whether this bounding box is a car or not by passing through VGG
#remove bounding box if below score threshold
#instantiates a correlation object with the boxes from the previous frame and this frame
self.corr = correlationClass(boundingBoxes, currentBoxes)
#correlates the bounding boxes. method to be implemented in correlation.py.
#tracked and new each contain a list of indices for the bounding boxes in this frame,
#whether the car in the box is matched with a bounding box in the previous frame, or not.
tracked, new = self.corr.correlateBoxes(prevImage, img)
for i in range(len(currentBoxes)):
(x1, x2, y1, y2) = currentBoxes[i]
if i in new:
img = self.drawBox(img, x1, x2, y1, y2, [0, 255, 0])
else:
img = self.drawBox(img, x1, x2, y1, y2, [0, 0, 255])
#img = self.filter(img)
print("Image 1 bounding boxes: " + str(len(boundingBoxes)))
print("Image 2 bounding boxes: " + str(len(currentBoxes)))
print("Number of correlations: " + str(self.corr.numCorrelations))
T.boundingBoxes = currentBoxes
T.total += len(new)
T.prevImage = img
font = cv2.FONT_HERSHEY_SIMPLEX
cv2.putText(img,'Car Count: ' + str(T.total),(10,230), font, 0.5,(255,255,255),2,cv2.LINE_AA)
retval, b = cv2.imencode('.jpg', img)
retval2, b2 = cv2.imencode('.jpg', img2)
encoded_string = base64.b64encode(b)
encoded_string2 = base64.b64encode(b2)
D = {
"im1":encoded_string,
"im2":encoded_string2
}
json_data = json.dumps(D)
#encoded_string = base64.b64encode(arr)
return json_data
def predict(self, image, image_type):
keras.backend.set_session(self.session)
import scipy.misc as sm
# image = cv2.resize(image, (self.WIDTH, self.HEIGHT))
ret, mat = cv2.imencode(".jpg", image)
image = Image(blob=mat)
image.resize(self.WIDTH, self.HEIGHT)
image.format = 'gray'
image.alpha_channel = False
image = np.asarray(bytearray(image.make_blob()),
dtype='uint8').reshape(image.size)
# if (self.both_model_flag):
# # image = cv2.resize(image, (self.WIDTH, self.HEIGHT), interpolation = cv2.INTER_NEAREST)
# # image = cv2.resize(image, (self.WIDTH, self.HEIGHT), interpolation = cv2.INTER_LINEAR)
# # image = sm.imresize(image,size=(self.WIDTH, self.HEIGHT),interp="bilinear")
# image = cv2.resize(image, (self.WIDTH, self.HEIGHT))
#
#
# else:
# # image = sm.imresize(image,size=(self.WIDTH, self.HEIGHT), interp="nearest")
# # image = cv2.resize(image, (self.WIDTH, self.HEIGHT), interpolation = cv2.INTER_LINEAR)
# # image = cv2.resize(image, (self.WIDTH, self.HEIGHT), interpolation = cv2.INTER_NEAREST)
# image = cv2.resize(image, (self.WIDTH, self.HEIGHT))
image = cv2.resize(image, (self.WIDTH, self.HEIGHT))
image = cv2.cvtColor(image, cv2.COLOR_GRAY2RGB) / 255
features = np.zeros(shape=(1, self.feature_model_output_shape[1],
self.feature_model_output_shape[2],
self.feature_model_output_shape[3]))
with self.graph.as_default():
with self.session.as_default():
features[0:1] = self.feature_model.predict(np.array([image]))
features = np.reshape(features,
(1, self.feature_model_output_shape[1] *
self.feature_model_output_shape[2] *
self.feature_model_output_shape[3]))
print(features[0][20])
if (self.both_model_flag):
with self.graph.as_default():
with self.session.as_default():
prediction = self.both_model.predict_classes(features)[0]
prob = self.both_model.predict(features)
pred_label = self.both_classes[prediction]
else:
if image_type == 1:
with self.graph.as_default():
with self.session.as_default():
prediction = self.letters_model.predict_classes(
features)[0]
prob = self.letters_model.predict(features)
pred_label = self.letters_classes[prediction]
elif image_type == 2:
with self.graph.as_default():
with self.session.as_default():
prediction = self.numbers_model.predict_classes(
features)[0]
prob = self.numbers_model.predict(features)
pred_label = self.numbers_classes[prediction]
else:
raise TypeError
probability = prob[0][prediction]
return pred_label, probability
obj = {}
obj['timestamp'] = datetime.now().isoformat()
obj['bytes'] = stringify_jpg(jpg_bytes)
return json.dumps(obj)
def deserialize_payload(payload):
return json.loads(payload)
#def deserialize_jpg(jpg_json):
# """Deserialized JSON object created by josnify_image.
#
# :param string :
# :return:
# :rtype:
# """
# return json.loads(jpg_json)
if __name__ == '__main__':
im = cv2.imread('/home/debug/codes/darknet/data/dog.jpg')
retval, jpg_bytes = cv2.imencode('.jpg', im)
# size of stringified dog.jpg is 1.33x larger than original
s_jpg = serialize_jpg(jpg_bytes)
d_jpg = deserialize_payload(s_jpg)
# TODO: Can we write JPEG bytes into file directly to prevent
# bytes -> numpy array -> decode RGB -> write encoded JPEG
cv2.imwrite('/tmp/dog.jpg', jpg2bgr(destringify_jpg(d_jpg['bytes'])))
def get_frame(self):
frame = self.flip_if_needed(self.vs.read())
ret, jpeg = cv2.imencode('.jpg', frame)
return jpeg.tobytes()
def save_full_sem_seg(self, cls_id, seg, group):
seg_str = cv2.imencode('.png', seg)[1]
key = 'label_semantic_segmentation_full_size/{:02d}'.format(cls_id)
self.save(key, seg_str, group)
self.send_response(200)
self.send_header("Content-type", "image/jpeg")
self.end_headers()
# ret, frame = cap.read()
# _, jpg = cv2.imencode(".jpg", frame)
self.wfile.write(jpg)
else:
self.send_response(404)
# continuously get frames from webcam
cap = cv2.VideoCapture(0)
ret, frame = cap.read()
_, jpg = cv2.imencode(".jpg", frame)
class FrameThread(threading.Thread):
def __init__(self):
threading.Thread.__init__(self)
self.isRunning = True
def run(self):
global jpg, cap
while self.isRunning:
ret, frame = cap.read()
_, jpg = cv2.imencode(".jpg", frame)
sleep(0.03)
def save_ori(self, ori, group):
ori_str = cv2.imencode('.png', ori)[1]
self.save('orientation', ori_str, group)
def save_seg(self, seg_id, seg, group):
seg_str = cv2.imencode('.png', seg)[1]
key = 'label_segmentation/{:02d}'.format(seg_id)
self.save(key, seg_str, group)
SECRET_KEY = 'Q6WtWCSFqqMAbKffCrqHQ6DDWRNc0oD7'
client = AipBodyAnalysis(APP_ID, API_KEY, SECRET_KEY)
parser = argparse.ArgumentParser()
parser.add_argument('-d', '--debug', type=bool, default=False)
args = parser.parse_args()
is_debug = args.debug
camera = PyK4A()
camera.config.color_resolution = ColorResolution.RES_1080P
camera.start()
c_type = 1
gesture = 0
print('starting...')
target_gesture = 3
while True:
capture = camera.get_capture()
img_bytes = cv2.imencode('.jpg', capture.color)[1].tobytes()
attr_thread = threading.Thread(target=deal_with_bodyAttr_result,
args=(img_bytes, ))
analysis_thread = threading.Thread(
target=deal_with_bodyAnalysis_result, args=(img_bytes, ))
attr_thread.start()
analysis_thread.start()
attr_thread.join()
analysis_thread.join()
final_results = integrate_results(attr_result, analysis_result)
if len(final_results) == 0:
continue
print(colored(final_results))
drawn_image = drawDebugImage(capture.color, final_results)
if is_debug:
cv2.imshow("Detected Image", drawn_image)
def save_full_image(self, img, group):
img_str = cv2.imencode('.png', img)[1]
self.save('input_full_size', img_str, group)
def jpeg_compression(image, factor):
encode_param = [int(cv2.IMWRITE_JPEG_QUALITY), factor]
_, image = cv2.imencode('.jpg', image, encode_param)
return cv2.imdecode(image, 1)
def frame_to_data(self, frame):
return cv2.imencode('.png', frame)[1].tobytes()
