def __init__(self, devid, device, model_xml, frameBuffer, results,
camera_width, camera_height, number_of_ncs, vidfps, nPoints,
w, h, new_w, new_h):
self.devid = devid
self.frameBuffer = frameBuffer
self.model_xml = model_xml
self.model_bin = os.path.splitext(model_xml)[0] + ".bin"
self.camera_width = camera_width
self.camera_height = camera_height
self.threshold = 0.1
self.nPoints = nPoints
self.num_requests = 4
self.inferred_request = [0] * self.num_requests
self.heap_request = []
self.inferred_cnt = 0
self.plugin = IEPlugin(device=device)
if "CPU" == device:
if platform.processor() == "x86_64":
self.plugin.add_cpu_extension("lib/libcpu_extension.so")
self.net = IENetwork(model=self.model_xml, weights=self.model_bin)
self.input_blob = next(iter(self.net.inputs))
self.exec_net = self.plugin.load(network=self.net,
num_requests=self.num_requests)
self.results = results
self.number_of_ncs = number_of_ncs
self.predict_async_time = 250
self.skip_frame = 0
self.roop_frame = 0
self.vidfps = vidfps
self.w = w #432
self.h = h #368
self.new_w = new_w
self.new_h = new_h
def __init__(self, devid, frameBuffer, results, camera_width,
camera_height, number_of_ncs, vidfps):
self.devid = devid
self.frameBuffer = frameBuffer
self.model_xml = model_xml
self.model_bin = model_bin
self.camera_width = camera_width
self.camera_height = camera_height
self.m_input_size = 256
self.threshould = 0.7
self.num_requests = 1
self.inferred_request = [0] * self.num_requests
self.heap_request = []
self.inferred_cnt = 0
self.plugin = IEPlugin(device="MYRIAD")
self.net = IENetwork(model=self.model_xml, weights=self.model_bin)
self.input_blob = next(iter(self.net.inputs))
self.out_blob = next(iter(self.net.outputs))
self.exec_net = self.plugin.load(network=self.net,
num_requests=self.num_requests)
self.results = results
self.number_of_ncs = number_of_ncs
self.predict_async_time = 800
self.skip_frame = 0
self.roop_frame = 0
self.vidfps = vidfps
self.new_w = int(camera_width * self.m_input_size / camera_width)
self.new_h = int(camera_height * self.m_input_size / camera_height)
def __init__(self, devid, frameBuffer, results, camera_width, camera_height, number_of_ncs, vidfps, api_results, model_name="coco_tiny_yolov3_320", input_size=320, plugin=None):
self.devid = devid
self.frameBuffer = frameBuffer
# self.model_xml = "./lrmodels/tiny-YoloV3/FP16/frozen_tiny_yolo_v3.xml"
# self.model_bin = "./lrmodels/tiny-YoloV3/FP16/frozen_tiny_yolo_v3.bin"
self.model_xml = "./cs6220-ncs2-yolo-api/models/%s.xml" % model_name
self.model_bin = "./cs6220-ncs2-yolo-api/models/%s.bin" % model_name
self.camera_width = camera_width
self.camera_height = camera_height
self.m_input_size = input_size
self.threshould = 0.4
self.num_requests = 4
self.inferred_request = [0] * self.num_requests
self.heap_request = []
self.inferred_cnt = 0
self.plugin = plugin
self.net = IENetwork(model=self.model_xml, weights=self.model_bin)
# self.net2 = IENetwork(model=self.model_xml, weights=self.model_bin)
self.input_blob = next(iter(self.net.inputs))
self.exec_net = self.plugin.load(network=self.net, num_requests=self.num_requests)
# self.exec_net2 = self.plugin.load(network=self.net2, num_requests=self.num_requests)
self.results = results
self.api_results = api_results
self.number_of_ncs = number_of_ncs
self.predict_async_time = 800
self.skip_frame = 0
self.roop_frame = 0
self.vidfps = vidfps
self.new_w = int(camera_width * min(self.m_input_size/camera_width, self.m_input_size/camera_height))
self.new_h = int(camera_height * min(self.m_input_size/camera_width, self.m_input_size/camera_height))
def __init__(self, devid, frameBuffer, results, camera_mode, camera_width,
camera_height, number_of_ncs, vidfps, skpfrm):
self.devid = devid
self.frameBuffer = frameBuffer
self.model_xml = "./lrmodel/MobileNetSSD/MobileNetSSD_deploy.xml"
self.model_bin = "./lrmodel/MobileNetSSD/MobileNetSSD_deploy.bin"
self.camera_width = camera_width
self.camera_height = camera_height
self.num_requests = 4
self.inferred_request = [0] * self.num_requests
self.heap_request = []
self.inferred_cnt = 0
self.plugin = IEPlugin(device="MYRIAD")
self.net = IENetwork(model=self.model_xml, weights=self.model_bin)
self.input_blob = next(iter(self.net.inputs))
self.exec_net = self.plugin.load(network=self.net,
num_requests=self.num_requests)
self.results = results
self.camera_mode = camera_mode
self.number_of_ncs = number_of_ncs
if self.camera_mode == 0:
self.skip_frame = skpfrm
else:
self.skip_frame = 0
self.roop_frame = 0
self.vidfps = vidfps
def __init__(self, devid, frameBuffer, results, camera_width, camera_height, number_of_ncs, vidfps):
self.devid = devid
self.frameBuffer = frameBuffer
self.model_xml = "./lrmodels/tiny-YoloV3/FP16/frozen_tiny_yolo_v3.xml"
self.model_bin = "./lrmodels/tiny-YoloV3/FP16/frozen_tiny_yolo_v3.bin"
self.camera_width = camera_width
self.camera_height = camera_height
self.m_input_size = 416
self.threshould = 0.4
self.num_requests = 4
self.inferred_request = [0] * self.num_requests
self.heap_request = []
self.inferred_cnt = 0
self.plugin = IEPlugin(device="MYRIAD")
self.net = IENetwork(model=self.model_xml, weights=self.model_bin)
self.input_blob = next(iter(self.net.inputs))
self.exec_net = self.plugin.load(network=self.net, num_requests=self.num_requests)
self.results = results
self.number_of_ncs = number_of_ncs
self.predict_async_time = 800
self.skip_frame = 0
self.roop_frame = 0
self.vidfps = vidfps
self.new_w = int(camera_width * min(self.m_input_size/camera_width, self.m_input_size/camera_height))
self.new_h = int(camera_height * min(self.m_input_size/camera_width, self.m_input_size/camera_height))
def __init__(self, config):
#Numbers to keep track of process ID
self.curr = 0
self.next = 1
self.width, self.height = config.frame_width, config.frame_height
try:
self.plugin = IEPlugin(device=config.default_device)
self.net = IENetwork(model=config.network_file,
weights=config.weights_file)
except RuntimeError:
print(
"We're probably dealing with a mac here, do some cpu target stuff"
)
self.plugin = IEPlugin(device=config.fallback_device)
self.net = IENetwork(model=config.network_file,
weights=config.weights_file)
supported_layers = self.plugin.get_supported_layers(self.net)
not_supported_layers = [
l for l in self.net.layers.keys() if l not in supported_layers
]
if len(not_supported_layers) != 0:
raise Exception(
"Some layers in the mdoel are not supported by the CPU - figure this out"
)
#Get sizes for image pre-processing
self.input_blob = next(iter(self.net.inputs))
self.output_blob = next(iter(self.net.outputs))
self.n, self.c, self.h, self.w = self.net.inputs[self.input_blob].shape
self.threshold = config.threshold
if config.mode == 'classify':
self.threshold = [config.threshold, config.non_threshold]
self.normalize = config.normalize
self.exec_net = self.plugin.load(network=self.net)
#Do we need this?
del self.net
def __init__(self, image_queue, results):
self.image_queue = image_queue
self.model_xml = "image_analysis/model/{}.xml".format(
config.MODEL_NAME)
self.model_bin = "image_analysis/model/{}.bin".format(
config.MODEL_NAME)
self.threshold = 0.4
self.num_requests = 2
self.inferred_request = [0] * self.num_requests
self.heap_request = []
self.inferred_cnt = 0
self.plugin = IEPlugin(device=config.DEVICE)
self.net = IENetwork(model=self.model_xml, weights=self.model_bin)
self.input_blob = next(iter(self.net.inputs))
self.exec_net = self.plugin.load(network=self.net,
num_requests=self.num_requests)
self.results = results
def __init__(self, devid, model_path, number_of_ncs):
global g_plugin
global g_inferred_request
global g_heap_request
global g_inferred_cnt
global g_number_of_allocated_ncs
self.devid = devid
if number_of_ncs == 0:
self.num_requests = 4
elif number_of_ncs == 1:
self.num_requests = 4
elif number_of_ncs == 2:
self.num_requests = 2
elif number_of_ncs >= 3:
self.num_requests = 1
print("g_number_of_allocated_ncs =", g_number_of_allocated_ncs,
"number_of_ncs =", number_of_ncs)
if g_number_of_allocated_ncs < 1:
self.plugin = IEPlugin(device="CPU")
self.plugin.add_cpu_extension("./lib/libcpu_extension.so")
self.inferred_request = [0] * self.num_requests
self.heap_request = []
self.inferred_cnt = 0
g_plugin = self.plugin
g_inferred_request = self.inferred_request
g_heap_request = self.heap_request
g_inferred_cnt = self.inferred_cnt
g_number_of_allocated_ncs += 1
else:
self.plugin = g_plugin
self.inferred_request = g_inferred_request
self.heap_request = g_heap_request
self.inferred_cnt = g_inferred_cnt
self.model_xml = model_path + ".xml"
self.model_bin = model_path + ".bin"
self.net = IENetwork(model=self.model_xml, weights=self.model_bin)
self.input_blob = next(iter(self.net.inputs))
self.exec_net = self.plugin.load(network=self.net,
num_requests=self.num_requests)
def classify_frame(inputQueue, outputQueue):
cur_request_id = 0
next_request_id = 1
model_xml = "models/MobileNetSSD_deploy.xml"
model_bin = "models/MobileNetSSD_deploy.bin"
plugin = IEPlugin(device="MYRIAD")
net = IENetwork(model=model_xml, weights=model_bin)
input_blob = next(iter(net.inputs))
out_blob = next(iter(net.outputs))
exec_net = plugin.load(network=net, num_requests=2)
n, c, h, w = net.inputs[input_blob].shape
del net
while True:
if not inputQueue.empty():
frame = inputQueue.get()
in_frame = cv2.dnn.blobFromImage(cv2.resize(frame, (300, 300)),
0.007843, (300, 300), 127.5)
exec_net.start_async(request_id=next_request_id,
inputs={input_blob: in_frame})
if exec_net.requests[cur_request_id].wait(-1) == 0:
detections = exec_net.requests[cur_request_id].outputs[
out_blob]
data_out = []
for i in np.arange(0, detections.shape[2]):
inference = []
confidence = detections[0, 0, i, 2]
if confidence > 0.2:
idx = int(detections[0, 0, i, 1])
box = detections[0, 0, i, 3:7] * np.array(
[frameWidth, frameHeight, frameWidth, frameHeight])
(startX, startY, endX, endY) = box.astype("int")
inference.extend(
(idx, confidence, startX, startY, endX, endY))
data_out.append(inference)
outputQueue.put(data_out)
cur_request_id, next_request_id = next_request_id, cur_request_id
model_xml = "models/train/test/openvino/mobilenet_v2_0.5_224/FP32/frozen-model.xml"
elif "GPU" == args.device or "MYRIAD" == args.device:
if args.boost == False:
model_xml = "models/train/test/openvino/mobilenet_v2_1.4_224/FP16/frozen-model.xml"
else:
model_xml = "models/train/test/openvino/mobilenet_v2_0.5_224/FP16/frozen-model.xml"
else:
print(
"Specify the target device to infer on; CPU, GPU, MYRIAD is acceptable."
)
sys.exit(0)
model_bin = os.path.splitext(model_xml)[0] + ".bin"
net = IENetwork(model=model_xml, weights=model_bin)
input_blob = next(iter(net.inputs))
exec_net = plugin.load(network=net)
inputs = net.inputs["image"]
h = inputs.shape[2] #368
w = inputs.shape[3] #432
threshold = 0.1
nPoints = 18
try:
while True:
t1 = time.perf_counter()
formatter = logging.Formatter('%(asctime)s [%(levelname)s] %(message)s')
stream_hander = logging.StreamHandler()
stream_hander.setFormatter(formatter)
logger.addHandler(stream_hander)
log_max_size = 10*1024*1024
log_file_count = 20
file_handler = handlers.RotatingFileHandler(filename = "log.log", maxBytes = log_max_size, backupCount=log_file_count)
file_handler.setFormatter(formatter)
logger.addHandler(file_handler)
# network 생성
logger.info("Network generation")
net = IENetwork(model = xml_path,weights = bin_path)
logger.info("Network generation Success")
# device (MYRIAD : NCS2)
logger.info("Device Init")
plugin = IEPlugin(device='MYRIAD')
logger.info("Device Init Success")
logger.info("Network Load...")
exec_net = plugin.load(net)
logger.info("Network Load Success")
def filtering_box(objects):
# Filtering overlapping boxes
# box를 걸러낸다
objlen = len(objects)
def main_IE_infer():
fps = ""
framepos = 0
frame_count = 0
vidfps = 0
skip_frame = 0
elapsedTime = 0
args = build_argparser().parse_args()
#model_xml = "lrmodels/YoloV3/FP32/frozen_yolo_v3.xml" #<--- CPU
model_xml = "/home/saket/openvino_models/tiger/n_yolov3_tiger_13.xml" #<--- MYRIAD
model_bin = os.path.splitext(model_xml)[0] + ".bin"
'''
cap = cv2.VideoCapture(0)
cap.set(cv2.CAP_PROP_FPS, 30)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, camera_width)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, camera_height)
'''
#cap = cv2.VideoCapture("data/input/testvideo.mp4")
#camera_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
#camera_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
#frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
#vidfps = int(cap.get(cv2.CAP_PROP_FPS))
#print("videosFrameCount =", str(frame_count))
#print("videosFPS =", str(vidfps))
time.sleep(1)
plugin = IEPlugin(device=args.device)
if "CPU" in args.device:
print('executing in cpu')
plugin.add_cpu_extension("lib/libcpu_extension.so")
net = IENetwork(model=model_xml, weights=model_bin)
input_blob = next(iter(net.inputs))
exec_net = plugin.load(network=net)
counter = 0
import glob
images = glob.glob(
'/media/saket/014178da-fdf2-462c-b901-d5f4dbce2e275/nn/PyTorch-YOLOv3/data/tigersamples/*.jpg'
)[0:20]
print('Num images', len(images))
from PIL import Image
while cv2.waitKey(0):
if cv2.waitKey(1) & 0xFF == ord('q') or counter > len(images) - 1:
break
t1 = time.time()
## Uncomment only when playing video files
#cap.set(cv2.CAP_PROP_POS_FRAMES, framepos)
image = np.asarray(Image.open(images[counter]).convert('RGB'))
#new_img = image
#image=cv2.imread(images[counter])
#image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
#image = np.asarray(Image.open(images[counter]).convert('RGB'))
#camera_height, camera_width = image.shape[0:2]
master = np.zeros((1920, 1920, 3), dtype='uint8')
master[420:1500, :] = image
image = master
camera_height, camera_width = image.shape[0:2]
#prepimg = image.transpose((2,0,1))
#prepimg, _ = pad_to_square(prepimg, 0)
# Resize
#prepimg = resize(prepimg, m_input_size)[np.newaxis]
#print (prepimg.shape)
#print (camera_height,camera_width)
#new_w = int(camera_width * min(m_input_size/camera_width, m_input_size/camera_height))
#new_h = int(camera_height * min(m_input_size/camera_width, m_input_size/camera_height))
#print (new_w,new_h)
new_w, new_h = 416, 416
print(new_w, new_h)
resized_image = cv2.resize(image, (new_w, new_h),
interpolation=cv2.INTER_CUBIC)
print('resized', resized_image.shape)
canvas = np.full((m_input_size, m_input_size, 3), 128)
canvas[(m_input_size - new_h) // 2:(m_input_size - new_h) // 2 + new_h,
(m_input_size - new_w) // 2:(m_input_size - new_w) // 2 +
new_w, :] = resized_image
prepimg = canvas
#resized_image = cv2.resize(canvas, (m_input_size, m_input_size), interpolation = cv2.INTER_CUBIC)
#print (resized_image.dtype)
#prepimg = resized_image.astype('float32')
#new_h, new_w = 416, 416
#master = np.zeros((1920,1920,3))
#master[420:1500,:,:]=new_img
#prepimg = cv2.resize(master, (416, 416), interpolation = cv2.INTER_CUBIC)
prepimg = prepimg[np.newaxis, :, :, :] # Batch size axis add
prepimg = prepimg.transpose((0, 3, 1, 2)) # NHWC to NCHW
print(prepimg.shape)
outputs = exec_net.infer(inputs={input_blob: prepimg})
objects = []
for output in outputs.values():
objects = ParseYOLOV3Output(output, new_h, new_w, camera_height,
camera_width, 0.5, objects)
# Filtering overlapping boxes
objlen = len(objects)
for i in range(objlen):
if (objects[i].confidence == 0.0):
continue
for j in range(i + 1, objlen):
if (IntersectionOverUnion(objects[i], objects[j]) >= 0.5):
objects[j].confidence = 0
# Drawing boxes
image = image[420:1500, :]
print(image.shape)
for obj in objects:
if obj.confidence < 0.2:
continue
label = obj.class_id
confidence = obj.confidence
if confidence > 0.2:
label_text = LABELS[label] + " (" + "{:.1f}".format(
confidence * 100) + "%)"
cv2.rectangle(image, (obj.xmin, obj.ymin - 420),
(obj.xmax, obj.ymax - 420), box_color,
box_thickness)
cv2.putText(image, label_text, (obj.xmin, obj.ymin - 5),
cv2.FONT_HERSHEY_SIMPLEX, 0.6, label_text_color, 1)
cv2.putText(image, fps, (camera_width - 170, 15),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (38, 0, 255), 1,
cv2.LINE_AA)
cv2.imshow("Result", image)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
elapsedTime = time.time() - t1
fps = "(Playback) {:.1f} FPS".format(1 / elapsedTime)
counter += 1
## frame skip, video file only
#skip_frame = int((vidfps - int(1/elapsedTime)) / int(1/elapsedTime))
#framepos += skip_frame
cv2.destroyAllWindows()
'''
img_detections = [] # Stores detections for each image index
itime=[]
for batch_i, input_imgs in enumerate(images):
# Configure input
input_imgs = np.asarray(Image.open(input_imgs).convert('RGB'))
input_imgs = input_imgs.transpose((2,0,1))
prepimg, _ = pad_to_square(input_imgs, 0)
# Resize
prepimg = resize(prepimg, m_input_size)[np.newaxis]
print (prepimg.shape)
# Get detections
prev_time = time.time()
detections = exec_net.infer(inputs={input_blob: prepimg})
detections = list(detections.values())
print (len(detections),len(detections[0]))
detections = non_max_suppression(detections, 0.5,0.5)
# Log progress
current_time = time.time()
inference_time = current_time - prev_time
itime.append(inference_time)
prev_time = current_time
#flops, params = profile(model, inputs=(input_imgs, ))
print("\t+ Batch %d, Inference Time: %.4f" % (batch_i, inference_time))
img_detections.extend(detections)
'''
del net
del exec_net
del plugin
def main_IE_infer():
camera_width = 1280
camera_height = 960
fps = ""
framepos = 0
frame_count = 0
vidfps = 0
skip_frame = 0
elapsedTime = 0
new_w = int(camera_width * min(m_input_size/camera_width, m_input_size/camera_height))
new_h = int(camera_height * min(m_input_size/camera_width, m_input_size/camera_height))
args = build_argparser().parse_args()
#model_xml = "lrmodels/tiny-YoloV3/FP32/frozen_tiny_yolo_v3.xml" #<--- CPU
#model_xml = "lrmodels/tiny-YoloV3/FP16/frozen_tiny_yolo_v3.xml" #<--- MYRIAD
model_xml = "frozen_darknet_yolov3_tiny_model_signals_CPU.xml"
model_bin = os.path.splitext(model_xml)[0] + ".bin"
# cap = cv2.VideoCapture(0)
# cap.set(cv2.CAP_PROP_FPS, 30)
# cap.set(cv2.CAP_PROP_FRAME_WIDTH, camera_width)
# cap.set(cv2.CAP_PROP_FRAME_HEIGHT, camera_height)
#cap = cv2.VideoCapture("ableitung.mp4")
#camera_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
#camera_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
#frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
#vidfps = int(cap.get(cv2.CAP_PROP_FPS))
#print("videosFrameCount =", str(frame_count))
#print("videosFPS =", str(vidfps))
img = cv2.imread("signals-dataset/yolov3/signal-1516.jpg")
#cam = cv2.VideoCapture(0)
#s, img = cam.read()
time.sleep(1)
plugin = IEPlugin(device=args.device)
if "CPU" in args.device:
plugin.add_cpu_extension("libcpu_extension_sse4.so")
net = IENetwork(model=model_xml, weights=model_bin)
input_blob = next(iter(net.inputs))
exec_net = plugin.load(network=net)
start_time = time.time()
resized_image = cv2.resize(img, (new_w, new_h), interpolation = cv2.INTER_CUBIC)
canvas = np.full((m_input_size, m_input_size, 3), 128)
canvas[(m_input_size-new_h)//2:(m_input_size-new_h)//2 + new_h,(m_input_size-new_w)//2:(m_input_size-new_w)//2 + new_w, :] = resized_image
prepimg = canvas
prepimg = prepimg[np.newaxis, :, :, :] # Batch size axis add
prepimg = prepimg.transpose((0, 3, 1, 2)) # NHWC to NCHW
outputs = exec_net.infer(inputs={input_blob: prepimg})
print('time taken: {}'.format(time.time() - start_time))
objects = []
for output in outputs.values():
objects = ParseYOLOV3Output(output, new_h, new_w, camera_height, camera_width, 0.4, objects)
for object in objects:
print(LABELS[object.class_id], object.confidence, "%")
# Filtering overlapping boxes
objlen = len(objects)
for i in range(objlen):
if (objects[i].confidence == 0.0):
continue
for j in range(i + 1, objlen):
if (IntersectionOverUnion(objects[i], objects[j]) >= 0.4):
if objects[i].confidence < objects[j].confidence:
objects[i], objects[j] = objects[j], objects[i]
objects[j].confidence = 0.0
# Drawing boxes
for obj in objects:
if obj.confidence < 0.02:
continue
label = obj.class_id
confidence = obj.confidence
#if confidence >= 0.2:
label_text = LABELS[label] + " (" + "{:.1f}".format(confidence * 100) + "%)"
cv2.rectangle(img, (obj.xmin, obj.ymin), (obj.xmax, obj.ymax), box_color, box_thickness)
cv2.putText(img, label_text, (obj.xmin, obj.ymin - 5), cv2.FONT_HERSHEY_SIMPLEX, 0.6, label_text_color, 1)
cv2.putText(img, fps, (camera_width - 170, 15), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (38, 0, 255), 1, cv2.LINE_AA)
cv2.imwrite("Result.jpg", img)
cv2.destroyAllWindows()
del net
del exec_net
del plugin
def main(camera_FPS, camera_width, camera_height, inference_scale, threshold,
device):
path = "pictures/"
if not os.path.exists(path):
os.mkdir(path)
model_path = "OneClassAnomalyDetection-RaspberryPi3/DOC/model/"
if os.path.exists(model_path):
# LOF
print("LOF model building...")
x_train = np.loadtxt(model_path + "train.csv", delimiter=",")
ms = MinMaxScaler()
x_train = ms.fit_transform(x_train)
# fit the LOF model
clf = LocalOutlierFactor(n_neighbors=5)
clf.fit(x_train)
# DOC
print("DOC Model loading...")
if device == "MYRIAD":
model_xml = "irmodels/tensorflow/FP16/weights.xml"
model_bin = "irmodels/tensorflow/FP16/weights.bin"
else:
model_xml = "irmodels/tensorflow/FP32/weights.xml"
model_bin = "irmodels/tensorflow/FP32/weights.bin"
net = IENetwork(model=model_xml, weights=model_bin)
plugin = IEPlugin(device=device)
if device == "CPU":
if platform.processor() == "x86_64":
plugin.add_cpu_extension("lib/x86_64/libcpu_extension.so")
exec_net = plugin.load(network=net)
input_blob = next(iter(net.inputs))
print("loading finish")
else:
print("Nothing model folder")
sys.exit(0)
base_range = min(camera_width, camera_height)
stretch_ratio = inference_scale / base_range
resize_image_width = int(camera_width * stretch_ratio)
resize_image_height = int(camera_height * stretch_ratio)
if base_range == camera_height:
crop_start_x = (resize_image_width - inference_scale) // 2
crop_start_y = 0
else:
crop_start_x = 0
crop_start_y = (resize_image_height - inference_scale) // 2
crop_end_x = crop_start_x + inference_scale
crop_end_y = crop_start_y + inference_scale
fps = ""
message = "Push [p] to take a picture"
result = "Push [s] to start anomaly detection"
flag_score = False
picture_num = 1
elapsedTime = 0
score = 0
score_mean = np.zeros(10)
mean_NO = 0
cap = cv2.VideoCapture(0)
cap.set(cv2.CAP_PROP_FPS, camera_FPS)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, camera_width)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, camera_height)
time.sleep(1)
while cap.isOpened():
t1 = time.time()
ret, image = cap.read()
if not ret:
break
image_copy = image.copy()
# prediction
if flag_score == True:
prepimg = cv2.resize(image,
(resize_image_width, resize_image_height))
prepimg = prepimg[crop_start_y:crop_end_y, crop_start_x:crop_end_x]
prepimg = np.array(prepimg).reshape(
(1, inference_scale, inference_scale, 3))
prepimg = prepimg / 255
prepimg = prepimg.transpose((0, 3, 1, 2))
exec_net.start_async(request_id=0, inputs={input_blob: prepimg})
exec_net.requests[0].wait(-1)
outputs = exec_net.requests[0].outputs["Reshape_"]
outputs = outputs.reshape((len(outputs), -1))
outputs = ms.transform(outputs)
score = -clf._decision_function(outputs)
# output score
if flag_score == False:
cv2.putText(image, result, (camera_width - 350, 100),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 1,
cv2.LINE_AA)
else:
score_mean[mean_NO] = score[0]
mean_NO += 1
if mean_NO == len(score_mean):
mean_NO = 0
if np.mean(score_mean) > threshold: #red if score is big
cv2.putText(image, "{:.1f} Score".format(np.mean(score_mean)),
(camera_width - 230, 100),
cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 0, 255), 1,
cv2.LINE_AA)
else: # blue if score is small
cv2.putText(image, "{:.1f} Score".format(np.mean(score_mean)),
(camera_width - 230, 100),
cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 1,
cv2.LINE_AA)
# message
cv2.putText(image, message, (camera_width - 285, 15),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 0), 1, cv2.LINE_AA)
cv2.putText(image, fps, (camera_width - 164, 50),
cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 0, 0), 1, cv2.LINE_AA)
cv2.imshow("Result", image)
# FPS
elapsedTime = time.time() - t1
fps = "{:.0f} FPS".format(1 / elapsedTime)
# quit or calculate score or take a picture
key = cv2.waitKey(1) & 0xFF
if key == ord("q"):
break
if key == ord("p"):
cv2.imwrite(path + str(picture_num) + ".jpg", image_copy)
picture_num += 1
if key == ord("s"):
flag_score = True
cv2.destroyAllWindows()
def camThread():
plugin = IEPlugin(device="CPU")
plugin.add_cpu_extension("./lib/libcpu_extension.so")
print("successful")
model_xml = "./models/face-detection-retail-0004.xml"
model_bin = "./models/face-detection-retail-0004.bin"
net = IENetwork(model=model_xml, weights=model_bin)
input_blob = next(iter(net.inputs))
out_blob = next(iter(net.outputs))
print("test")
Exec_net = plugin.load(network=net)
print("successful also")
cap = cv2.VideoCapture(0)
cv2.namedWindow('frame')
i = 0
while (i == i):
faces = []
ret, frame = cap.read()
if ret:
#frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
prepimg = cv2.resize(frame, (300, 300))
prepimg = prepimg[np.newaxis, :, :, :] # Batch size axis add
prepimg = prepimg.transpose((0, 3, 1, 2)) # NHWC to NCHW
outputs = Exec_net.infer(inputs={input_blob: prepimg})
for k, v in outputs.items():
print(str(k) + ' is key to value ' + str(v))
print(type(v))
print(v.flatten())
print(type(v.flatten()))
j = 0
while float(v.flatten()[j + 2]) > .9:
image_id = v.flatten()[j]
label = v.flatten()[j + 1]
conf = v.flatten()[j + 2]
boxTopLeftX = v.flatten()[j + 3] * frame.shape[1]
boxTopLefty = v.flatten()[j + 4] * frame.shape[0]
boxBottomRightX = v.flatten()[j + 5] * frame.shape[1]
boxBottomRightY = v.flatten()[j + 6] * frame.shape[0]
faces.append([
conf, (int(boxTopLeftX), int(boxTopLefty)),
(int(boxBottomRightX), int(boxBottomRightY))
])
j = j + 7
for face in faces:
frame = cv2.rectangle(np.array(frame), face[1], face[2],
(0, 0, 255), 2)
frame = cv2.putText(frame, "confidence: " + str(face[0]),
(face[1][0], face[1][1] - 2),
cv2.FONT_HERSHEY_SIMPLEX, .5,
(0, 0, 255), 2)
print(str(face[0]))
print("show image")
cv2.imshow('frame', frame)
#print("show frame")
if cv2.waitKey(1) & 0xFF == ord('q'):
break
i = i + 1
cap.release()
cv2.destroyAllWindows()
def main_IE_infer():
fps = ""
framepos = 0
frame_count = 0
vidfps = 0
skip_frame = 0
elapsedTime = 0
args = build_argparser().parse_args()
result_file = args.result
model_xml = args.model
test_dir = args.test_dir
model_bin = os.path.splitext(model_xml)[0] + ".bin"
time.sleep(1)
plugin = IEPlugin(device=args.device)
if "CPU" in args.device:
print('executing in cpu')
plugin.add_cpu_extension("lib/libcpu_extension.so")
net = IENetwork(model=model_xml, weights=model_bin)
input_blob = next(iter(net.inputs))
exec_net = plugin.load(network=net)
counter = 0
images = glob.glob(test_dir + '/*.jpg')
print('Num images', len(images))
from PIL import Image
results = []
for item in images:
t1 = time.time()
image = np.asarray(Image.open(item).convert('RGB'))
master = np.zeros((1920, 1920, 3), dtype='uint8')
master[420:1500, :] = image
image = master
camera_height, camera_width = image.shape[0:2]
new_w, new_h = 416, 416
resized_image = cv2.resize(image, (new_w, new_h),
interpolation=cv2.INTER_CUBIC)
canvas = np.full((m_input_size, m_input_size, 3), 0)
canvas[(m_input_size - new_h) // 2:(m_input_size - new_h) // 2 + new_h,
(m_input_size - new_w) // 2:(m_input_size - new_w) // 2 +
new_w, :] = resized_image
prepimg = canvas
prepimg = prepimg[np.newaxis, :, :, :] # Batch size axis add
prepimg = prepimg.transpose((0, 3, 1, 2)) # NHWC to NCHW
#print (prepimg.shape)
outputs = exec_net.infer(inputs={input_blob: prepimg})
objects = []
for output in outputs.values():
objects = ParseYOLOV3Output(output, new_h, new_w, camera_height,
camera_width, 0.5, objects)
# Filtering overlapping boxes
objlen = len(objects)
for i in range(objlen):
if (objects[i].confidence == 0.0):
continue
for j in range(i + 1, objlen):
if (IntersectionOverUnion(objects[i], objects[j]) >= 0.5):
objects[j].confidence = 0
# Save detection
for obj in objects:
if obj.confidence < 0.5:
continue
label = obj.class_id
confidence = obj.confidence
if confidence > 0.5:
xmin, ymin, xmax, ymax = obj.xmin, obj.ymin - 420, obj.xmax, obj.ymax - 420
x, y, w, h = xmin, ymin, xmax - xmin, ymax - ymin
anns = {
'image_id': item.split('/')[-1].replace('.jpg', ''),
'category_id': 1,
'segmentation': [],
'area': float(w * h),
'bbox': [x, y, w, h],
'score': float(confidence),
'iscrowd': 0
}
results.append(anns)
elapsedTime = time.time() - t1
fps = "(Playback) {:.1f} FPS".format(1 / elapsedTime)
print(fps)
json.dump(results, open(result_file, 'w'))
del net
del exec_net
del plugin
def main_IE_infer():
camera_width = 320
camera_height = 240
fps = ""
framepos = 0
frame_count = 0
vidfps = 0
skip_frame = 0
elapsedTime = 0
new_w = int(camera_width * m_input_size/camera_width)
new_h = int(camera_height * m_input_size/camera_height)
args = build_argparser().parse_args()
model_xml = "lrmodels/YoloV3/FP32/frozen_yolo_v3.xml" #<--- CPU
#model_xml = "lrmodels/YoloV3/FP16/frozen_yolo_v3.xml" #<--- MYRIAD
model_bin = os.path.splitext(model_xml)[0] + ".bin"
cap = cv2.VideoCapture(0)
cap.set(cv2.CAP_PROP_FPS, 30)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, camera_width)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, camera_height)
#cap = cv2.VideoCapture("data/input/testvideo.mp4")
#camera_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
#camera_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
#frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
#vidfps = int(cap.get(cv2.CAP_PROP_FPS))
#print("videosFrameCount =", str(frame_count))
#print("videosFPS =", str(vidfps))
time.sleep(1)
plugin = IEPlugin(device=args.device)
if "CPU" in args.device:
plugin.add_cpu_extension("lib/libcpu_extension.so")
net = IENetwork(model=model_xml, weights=model_bin)
input_blob = next(iter(net.inputs))
exec_net = plugin.load(network=net)
while cap.isOpened():
t1 = time.time()
## Uncomment only when playing video files
#cap.set(cv2.CAP_PROP_POS_FRAMES, framepos)
ret, image = cap.read()
if not ret:
break
resized_image = cv2.resize(image, (new_w, new_h), interpolation = cv2.INTER_CUBIC)
canvas = np.full((m_input_size, m_input_size, 3), 128)
canvas[(m_input_size-new_h)//2:(m_input_size-new_h)//2 + new_h,(m_input_size-new_w)//2:(m_input_size-new_w)//2 + new_w, :] = resized_image
prepimg = canvas
prepimg = prepimg[np.newaxis, :, :, :] # Batch size axis add
prepimg = prepimg.transpose((0, 3, 1, 2)) # NHWC to NCHW
outputs = exec_net.infer(inputs={input_blob: prepimg})
objects = []
for output in outputs.values():
objects = ParseYOLOV3Output(output, new_h, new_w, camera_height, camera_width, 0.7, objects)
# Filtering overlapping boxes
objlen = len(objects)
for i in range(objlen):
if (objects[i].confidence == 0.0):
continue
for j in range(i + 1, objlen):
if (IntersectionOverUnion(objects[i], objects[j]) >= 0.4):
objects[j].confidence = 0
# Drawing boxes
for obj in objects:
if obj.confidence < 0.2:
continue
label = obj.class_id
confidence = obj.confidence
if confidence > 0.2:
label_text = LABELS[label] + " (" + "{:.1f}".format(confidence * 100) + "%)"
cv2.rectangle(image, (obj.xmin, obj.ymin), (obj.xmax, obj.ymax), box_color, box_thickness)
cv2.putText(image, label_text, (obj.xmin, obj.ymin - 5), cv2.FONT_HERSHEY_SIMPLEX, 0.6, label_text_color, 1)
cv2.putText(image, fps, (camera_width - 170, 15), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (38, 0, 255), 1, cv2.LINE_AA)
cv2.imshow("Result", image)
if cv2.waitKey(1)&0xFF == ord('q'):
break
elapsedTime = time.time() - t1
fps = "(Playback) {:.1f} FPS".format(1/elapsedTime)
## frame skip, video file only
#skip_frame = int((vidfps - int(1/elapsedTime)) / int(1/elapsedTime))
#framepos += skip_frame
cv2.destroyAllWindows()
del net
del exec_net
del plugin
class_names = [
'Angry', 'Disgust', 'Fear', 'Happy', 'Sad', 'Surprise', 'Neutral'
]
t_prediction = 0
image = cv2.imread('/home/pi/Desktop/Expression5/1.jpg')
img = cv2.resize(image.astype(np.float32), (44, 44))
img -= imagenet_mean
img = img.reshape((1, 44, 44, 3))
img = img.transpose((0, 3, 1, 2))
model_xml_CPU = '/home/pi/Desktop/Expression5/models/torch_model.xml'
model_bin_CPU = '/home/pi/Desktop/Expression5/models/torch_model.bin'
plugin = IEPlugin(device='MYRIAD')
net = IENetwork(model=model_xml_CPU, weights=model_bin_CPU)
net.batch_size = 1
input_blob = next(iter(net.inputs))
exec_net = plugin.load(network=net)
t = time.time()
outputs = exec_net.infer(inputs={input_blob: img})
t_prediction = time.time() - t
class_name = class_names[np.argmax(outputs['617'])]
probs = outputs['617'][0, np.argmax(outputs['617'])]
print('Prediction time: %.2f' % t_prediction + ', Speed : %.2fFPS' %
(1 / t_prediction))
print("Class: " + class_name + ", probability: %.4f" % probs)
def main_IE_infer():
camera_width = 320
camera_height = 240
m_input_size = 513
fps = ""
framepos = 0
frame_count = 0
vidfps = 0
skip_frame = 0
elapsedTime = 0
args = build_argparser().parse_args()
#model_xml = "lrmodels/PascalVOC/FP32/frozen_inference_graph.xml" #<--- CPU
model_xml = "lrmodels/PascalVOC/FP16/frozen_inference_graph.xml" #<--- MYRIAD
model_bin = os.path.splitext(model_xml)[0] + ".bin"
seg_image = Image.open("data/input/009649.png")
palette = seg_image.getpalette() # Get a color palette
cap = cv2.VideoCapture(0)
cap.set(cv2.CAP_PROP_FPS, 10)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, camera_width)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, camera_height)
#cap = cv2.VideoCapture("data/input/testvideo.mp4")
#camera_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
#camera_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
#frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
#vidfps = int(cap.get(cv2.CAP_PROP_FPS))
#print("videosFrameCount =", str(frame_count))
#print("videosFPS =", str(vidfps))
time.sleep(1)
plugin = IEPlugin(device=args.device, plugin_dirs=args.plugin_dir)
if "CPU" in args.device:
plugin.add_cpu_extension("lib/libcpu_extension.so")
if args.performance:
plugin.set_config({"PERF_COUNT": "YES"})
# Read IR
net = IENetwork(model=model_xml, weights=model_bin)
input_blob = next(iter(net.inputs))
exec_net = plugin.load(network=net)
while cap.isOpened():
t1 = time.time()
#cap.set(cv2.CAP_PROP_POS_FRAMES, framepos) # Uncomment only when playing video files
ret, image = cap.read()
if not ret:
break
ratio = 1.0 * m_input_size / max(image.shape[0], image.shape[1])
shrink_size = (int(ratio * image.shape[1]),
int(ratio * image.shape[0]))
image = cv2.resize(image, shrink_size, interpolation=cv2.INTER_CUBIC)
prepimg = _pre._pre_process(image)
prepimg = prepimg.transpose((0, 3, 1, 2)) #NHWC to NCHW
res = exec_net.infer(inputs={input_blob: prepimg})
result = _post._post_process(res["ArgMax/Squeeze"], image)[0]
outputimg = Image.fromarray(np.uint8(result), mode="P")
outputimg.putpalette(palette)
outputimg = outputimg.convert("RGB")
outputimg = np.asarray(outputimg)
outputimg = cv2.cvtColor(outputimg, cv2.COLOR_RGB2BGR)
outputimg = cv2.addWeighted(image, 1.0, outputimg, 0.9, 0)
outputimg = cv2.resize(outputimg, (camera_width, camera_height))
cv2.putText(outputimg, fps, (camera_width - 180, 15),
cv2.FONT_HERSHEY_SIMPLEX, 0.5, (38, 0, 255), 1,
cv2.LINE_AA)
cv2.imshow("Result", outputimg)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
elapsedTime = time.time() - t1
fps = "(Playback) {:.1f} FPS".format(1 / elapsedTime)
# frame skip, video file only
skip_frame = int(
(vidfps - int(1 / elapsedTime)) / int(1 / elapsedTime))
framepos += skip_frame
cv2.destroyAllWindows()
del net
del exec_net
del plugin
def movidius(frameCount):
global outputFrame, lock, _reset
ct = CentroidTracker()
args = argsparser()
#ใช้ Log แทน Print
log.basicConfig(format="[ %(levelname)s ] %(message)s",
level=log.INFO,
stream=sys.stdout)
#Funtion Select
cam_select, cam_rtsp, min_conf_threshold, notify, detect_list, resW, resH = select_option(
)
# warmup
if cam_select == 0:
vs = cv2.VideoCapture(0)
else:
vs = cv2.VideoCapture(cam_rtsp)
imW, imH = int(resW), int(resH)
vs.set(3, imW)
vs.set(4, imH)
time.sleep(0.2)
##################
#โหลดโมดูล
model_xml = args.model
model_bin = os.path.splitext(model_xml)[0] + ".bin"
log.info("Initializing plugin for {} device...".format(args.device))
plugin = IEPlugin(device=args.device, plugin_dirs=args.plugin_dir)
if args.cpu_extension and 'CPU' in args.device:
plugin.add_cpu_extension('args.cpu_extension')
# Read IR
log.info("Reading IR...")
net = IENetwork(model=model_xml, weights=model_bin)
if plugin.device == "CPU":
supported_layers = plugin.get_supported_layers(net)
not_supported_layers = [
l for l in net.layers.keys() if l not in supported_layers
]
if len(not_supported_layers) != 0:
log.error(
"Following layers are not supported by the plugin for specified device {}:\n {}"
.format(plugin.device, ', '.join(not_supported_layers)))
log.error(
"Please try to specify cpu extensions library path in demo's command line parameters using -l "
"or --cpu_extension command line argument")
sys.exit(1)
assert len(
net.inputs.keys()) == 1, "Demo supports only single input topologies"
assert len(net.outputs) == 1, "Demo supports only single output topologies"
input_blob = next(iter(net.inputs))
out_blob = next(iter(net.outputs))
log.info("Loading IR to the plugin...")
exec_net = plugin.load(network=net, num_requests=2)
# Read and pre-process input image
n, c, h, w = net.inputs[input_blob].shape
del net
if args.labels:
with open(args.labels, 'r') as f:
labels_map = [x.strip() for x in f]
else:
labels_map = None
cur_request_id = 0
next_request_id = 1
log.info("Starting inference in async mode...")
log.info("To switch between sync and async modes press Tab button")
log.info("To stop the demo execution press Esc button")
is_async_mode = False
##################
# Initialize frame rate calculation
frame_rate_calc = 1
freq = cv2.getTickFrequency()
font = cv2.FONT_HERSHEY_SIMPLEX
##########################################
objects_first = 0
t1_image = time.time()
t2_image = time.time()
##########################################
log.info("Detect On")
while True:
t1 = cv2.getTickCount()
#ชื่อคลาสและไอดีของคลาส
ckname = ''
object_name = None
ret, frame_q = vs.read()
if is_async_mode:
in_frame = cv2.resize(frame_q, (w, h))
in_frame = in_frame.transpose(
(2, 0, 1)) # Change data layout from HWC to CHW
in_frame = in_frame.reshape((n, c, h, w))
exec_net.start_async(request_id=next_request_id,
inputs={input_blob: in_frame})
else:
in_frame = cv2.resize(frame_q, (w, h))
in_frame = in_frame.transpose(
(2, 0, 1)) # Change data layout from HWC to CHW
in_frame = in_frame.reshape((n, c, h, w))
exec_net.start_async(request_id=cur_request_id,
inputs={input_blob: in_frame})
rects = []
if exec_net.requests[cur_request_id].wait(-1) == 0:
# Parse detection results of the current request
res = exec_net.requests[cur_request_id].outputs[out_blob]
for obj in res[0][0]:
# Draw only objects when probability more than specified threshold
if obj[2] > min_conf_threshold:
# Draw label
class_id = int(obj[1])
object_name = labels_map[class_id] if labels_map else str(
class_id)
if object_name in detect_list:
ckname = ckname + object_name + ','
xmin = int(obj[3] * imW)
ymin = int(obj[4] * imH)
xmax = int(obj[5] * imW)
ymax = int(obj[6] * imH)
cv2.rectangle(frame_q, (xmin, ymin), (xmax, ymax),
(10, 255, 0), 2)
label = '%s: %s%%' % (object_name,
int(round(obj[2] * 100, 1)))
#print(label)
labelSize, baseLine = cv2.getTextSize(
label, cv2.FONT_HERSHEY_SIMPLEX, 0.7, 2)
label_ymin = max(ymin, labelSize[1] + 10)
color = color_name(object_name)
cv2.rectangle(
frame_q, (xmin, label_ymin - labelSize[1] - 10),
(xmin + labelSize[0], label_ymin + baseLine - 10),
color, cv2.FILLED)
cv2.putText(frame_q, label, (xmin, label_ymin - 7),
cv2.FONT_HERSHEY_SIMPLEX, 0.7, (0, 0, 0),
2)
x = np.array([xmin, ymin, xmax, ymax])
# print(x)
rects.append(x.astype("int"))
cv2.putText(frame_q, "FPS: {0:.2f}".format(frame_rate_calc), (30, 50),
font, 1, (255, 255, 0), 2, cv2.LINE_AA)
#Funtion display_show
t1_image = time.time()
frame_q, objects_first, t2_image = display_show(
rects, t1_image, t2_image, ckname, frame_q, objects_first, notify,
ct)
t2 = cv2.getTickCount()
time1 = (t2 - t1) / freq
frame_rate_calc = 1 / time1
if is_async_mode:
cur_request_id, next_request_id = next_request_id, cur_request_id
with lock:
outputFrame = frame_q.copy()
if _reset == 1:
sys.exit()
vs.stop()
def camThread():
plugin = IEPlugin(device="CPU")
plugin.add_cpu_extension("./lib/libcpu_extension.so")
print("successful")
model_xml = "./models/face-detection-retail-0004.xml"
model_bin = "./models/face-detection-retail-0004.bin"
net = IENetwork(model=model_xml, weights=model_bin)
input_blob = next(iter(net.inputs))
out_blob = next(iter(net.outputs))
print("test")
Exec_net = plugin.load(network=net)
print("successful also")
emotion_model_xml = "./models/emotions-recognition-retail-0003.xml"
emotion_model_bin = "./models/emotions-recognition-retail-0003.bin"
emotionNet = IENetwork(model=emotion_model_xml, weights=emotion_model_bin)
emotion_input_blob = next(iter(net.inputs))
emotion_exec_net = plugin.load(network=emotionNet)
print("Successfully loaded emotion model")
cap = cv2.VideoCapture(0)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, 640)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 480)
cv2.namedWindow('frame')
i = 0
start = datetime.now()
fps = 0
while (i <= i):
faces = []
ret, frame = cap.read()
if ret:
#frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
prepimg = cv2.resize(frame, (300, 300))
prepimg = prepimg[np.newaxis, :, :, :] # Batch size axis add
prepimg = prepimg.transpose((0, 3, 1, 2)) # NHWC to NCHW
outputs = Exec_net.infer(inputs={input_blob: prepimg})
for k, v in outputs.items():
#print(str(k) + ' is key to value ' + str(v))
#print(type(v))
#print(v.flatten())
#print(type(v.flatten()))
j = 0
while float(v.flatten()[j + 2]) > .9:
image_id = v.flatten()[j]
label = v.flatten()[j + 1]
conf = v.flatten()[j + 2]
boxTopLeftX = v.flatten()[j + 3] * frame.shape[1]
boxTopLefty = v.flatten()[j + 4] * frame.shape[0]
boxBottomRightX = v.flatten()[j + 5] * frame.shape[1]
boxBottomRightY = v.flatten()[j + 6] * frame.shape[0]
faces.append([
conf, (int(boxTopLeftX), int(boxTopLefty)),
(int(boxBottomRightX), int(boxBottomRightY))
])
j = j + 7
for face in faces:
prepimg = frame[face[1][1]:face[2][1],
face[1][0]:face[2][0]]
#prepimg = cv2.cvtColor(prepimg, cv2.COLOR_RGB2BGR)
prepimg = cv2.resize(frame, (64, 64))
prepimg = prepimg[
np.newaxis, :, :, :] # Batch size axis add
prepimg = prepimg.transpose((0, 3, 1, 2)) # NHWC to NCHW
emotion_outputs = emotion_exec_net.infer(
inputs={emotion_input_blob: prepimg})
for keys, values in emotion_outputs.items():
print(type(values))
print("emotion key is " + str(keys) +
" and value is " + str(values.flatten()))
print(int(np.argmax(values.flatten())))
emotion = LABELS[int(np.argmax(values.flatten()))]
prob = float(values.flatten()[int(
np.argmax(values.flatten()))]) * 100
print(emotion)
frame = cv2.rectangle(np.array(frame), face[1], face[2],
(0, 0, 255), 1)
if prob > 60:
frame = cv2.putText(
frame,
emotion + " confidence: " + "%.2f" % prob + "%",
(face[1][0], face[1][1] - 2),
cv2.FONT_HERSHEY_SIMPLEX, .5, (0, 0, 255), 1,
cv2.LINE_AA)
#print(str(face[0]))
frame = cv2.putText(frame, "FPS: " + "%.2f" % fps, (0, 15),
cv2.FONT_HERSHEY_SIMPLEX, .5, (0, 0, 255), 1)
print("show image")
cv2.imshow('frame', frame)
#print("show frame")
if cv2.waitKey(1) & 0xFF == ord('q'):
break
i = i + 1
if i > 30:
fps = i / int((datetime.now() - start).total_seconds())
print(fps)
cap.release()
cv2.destroyAllWindows()
def camThread(device, number_of_camera, camera_width, camera_height,
number_of_ncs, video, precision):
if device == 'CPU':
plugin = IEPlugin(device="CPU")
if platform.system() == "Linux":
plugin.add_cpu_extension('./lib/libcpu_extension.so')
elif platform.system() == "Windows":
print(os.getcwd())
plugin.add_cpu_extension(os.getcwd() + '\\lib\\cpu_extension.dll')
print("successfully loaded CPU plugin")
if precision == "FP32":
model_xml = "./models/FP32/face-detection-retail-0004.xml"
model_bin = "./models/FP32/face-detection-retail-0004.bin"
emotion_model_xml = "./models/FP32/emotions-recognition-retail-0003.xml"
emotion_model_bin = "./models/FP32/emotions-recognition-retail-0003.bin"
elif precision == "INT8":
model_xml = "./models/INT8/face-detection-retail-0004.xml"
model_bin = "./models/INT8/face-detection-retail-0004.bin"
emotion_model_xml = "./models/INT8/emotions-recognition-retail-0003.xml"
emotion_model_bin = "./models/INT8/emotions-recognition-retail-0003.bin"
if device == "MYRIAD":
plugin = IEPlugin(device="MYRIAD")
print("Successfully loaded MYRIAD plugin")
model_xml = "./models/FP16/face-detection-retail-0004.xml"
model_bin = "./models/FP16/face-detection-retail-0004.bin"
emotion_model_xml = "./models/FP16/emotions-recognition-retail-0003.xml"
emotion_model_bin = "./models/FP16/emotions-recognition-retail-0003.bin"
net = IENetwork(model=model_xml, weights=model_bin)
input_blob = next(iter(net.inputs))
output_blob = next(iter(net.outputs))
Exec_net = plugin.load(network=net)
print("successfully loaded face model")
emotionNet = IENetwork(model=emotion_model_xml, weights=emotion_model_bin)
emotion_input_blob = next(iter(emotionNet.inputs))
emotion_output_blob = next(iter(emotionNet.outputs))
emotion_exec_net = plugin.load(network=emotionNet)
print("Successfully loaded emotion model")
if video == "":
cap = cv2.VideoCapture(number_of_camera)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, camera_width)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, camera_height)
else:
cap = cv2.VideoCapture(video)
cv2.namedWindow('frame')
i = 0
start = datetime.now()
fps = 0
while (cap.isOpened()):
faces = []
ret, frame = cap.read()
if ret:
# frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
prepimg = cv2.resize(frame, (300, 300))
prepimg = prepimg[np.newaxis, :, :, :] # Batch size axis add
prepimg = prepimg.transpose((0, 3, 1, 2)) # NHWC to NCHW
infer_request_handle = Exec_net.start_async(
request_id=0, inputs={input_blob: prepimg})
infer_status = infer_request_handle.wait()
outputs = infer_request_handle.outputs[output_blob]
outputs = outputs.flatten()
j = 0
while float(outputs[j + 2] > .9):
image_id = outputs[j]
label = outputs[j + 1]
conf = outputs[j + 2]
boxTopLeftX = outputs[j + 3] * frame.shape[1]
boxTopLeftY = outputs[j + 4] * frame.shape[0]
boxBottomRightX = outputs[j + 5] * frame.shape[1]
boxBottomRightY = outputs[j + 6] * frame.shape[0]
faces.append([
conf, (int(boxTopLeftX), int(boxTopLeftY)),
(int(boxBottomRightX), int(boxBottomRightY))
])
j = j + 7
#Loop through faces and async infer, adding each handler to a dict. Then go through dict and draw on frame.
emotion_results = []
request_handler = None
for index, face in enumerate(faces):
prepimg = frame[face[1][1]:face[2][1], face[1][0]:face[2][0]]
prepimg = cv2.resize(frame, (64, 64))
prepimg = prepimg[np.newaxis, :, :, :] # Batch size axis add
prepimg = prepimg.transpose((0, 3, 1, 2)) # NHWC to NCHW
if index > 0:
start_time = datetime.now()
print("Waiting")
emotion_outputs_status = request_handler[0].wait()
end_time = datetime.now()
print(end_time - start_time)
emotion_results.append(
(request_handler[0].outputs[emotion_output_blob].
flatten(), request_handler[1]))
request_handler = (emotion_exec_net.start_async(
request_id=0, inputs={emotion_input_blob: prepimg}), face)
print("started async infer")
if request_handler:
emotion_outputs_status = request_handler[0].wait()
emotion_results.append(
emotion_results.append(
(request_handler[0].outputs[emotion_output_blob].
flatten(), request_handler[1])))
for emotion_result in emotion_results:
if emotion_result != None:
emotion_probs = emotion_result[0]
face = emotion_result[1]
emotion = LABELS[int(np.argmax(emotion_probs))]
prob = float(emotion_probs[int(
np.argmax(emotion_probs))]) * 100
print(emotion)
neutral = emotion_probs[0]
happy = emotion_probs[1]
sad = emotion_probs[2]
surprise = emotion_probs[3]
anger = emotion_probs[4]
prob_scale_x = face[1][0] - 100
frame = cv2.rectangle(np.array(frame), face[1], face[2],
(0, 0, 255), 1)
frame = cv2.rectangle(np.array(frame),
(face[1][0] - 100, face[1][1]),
(face[1][0], face[1][1] + 50),
(0, 0, 255), cv2.FILLED)
frame = cv2.rectangle(
np.array(frame), (prob_scale_x, face[1][1]),
(face[1][0] + int(neutral * 100 - 100),
face[1][1] + 10), (0, 255, 0), cv2.FILLED)
frame = cv2.putText(frame, "Neutral",
(prob_scale_x + 60, face[1][1] + 7),
cv2.FONT_HERSHEY_SIMPLEX, .3,
(0, 0, 0), 1)
frame = cv2.rectangle(
np.array(frame), (prob_scale_x, face[1][1] + 10),
(face[1][0] + int(happy * 100 - 100), face[1][1] + 20),
(0, 255, 0), cv2.FILLED)
frame = cv2.putText(frame, "Happy",
(prob_scale_x + 60, face[1][1] + 17),
cv2.FONT_HERSHEY_SIMPLEX, .3,
(0, 0, 0), 1)
frame = cv2.rectangle(
np.array(frame), (prob_scale_x, face[1][1] + 20),
(face[1][0] + int(sad * 100 - 100), face[1][1] + 30),
(0, 255, 0), cv2.FILLED)
frame = cv2.putText(frame, "Sad",
(prob_scale_x + 60, face[1][1] + 27),
cv2.FONT_HERSHEY_SIMPLEX, .3,
(0, 0, 0), 1)
frame = cv2.rectangle(
np.array(frame), (prob_scale_x, face[1][1] + 30),
(face[1][0] + int(surprise * 100 - 100),
face[1][1] + 40), (0, 255, 0), cv2.FILLED)
frame = cv2.putText(frame, "Surprise",
(prob_scale_x + 60, face[1][1] + 37),
cv2.FONT_HERSHEY_SIMPLEX, .3,
(0, 0, 0), 1)
frame = cv2.rectangle(
np.array(frame), (prob_scale_x, face[1][1] + 40),
(face[1][0] + int(anger * 100 - 100), face[1][1] + 50),
(0, 255, 0), cv2.FILLED)
frame = cv2.putText(frame, "Anger",
(prob_scale_x + 60, face[1][1] + 47),
cv2.FONT_HERSHEY_SIMPLEX, .3,
(0, 0, 0), 1)
if prob > 60:
frame = cv2.putText(
frame,
emotion + " confidence: " + "%.2f" % prob + "%",
(face[1][0], face[1][1] - 2),
cv2.FONT_HERSHEY_SIMPLEX, .5, (0, 0, 255), 1,
cv2.LINE_AA)
# print(str(face[0]))
frame = cv2.putText(frame, "FPS: " + "%.2f" % fps, (0, 15),
cv2.FONT_HERSHEY_SIMPLEX, .5, (0, 0, 255), 1)
print("show image")
cv2.imshow('frame', frame)
else:
break
# print("show frame")
if cv2.waitKey(1) & 0xFF == ord('q'):
break
i = i + 1
if int((datetime.now() - start).total_seconds()) > 0:
fps = i / int((datetime.now() - start).total_seconds())
print(fps)
print("Total time: " + str(datetime.now() - start))
cap.release()
cv2.destroyAllWindows()
def main_IE_infer():
fps = ""
framepos = 0
frame_count = 0
vidfps = 0
skip_frame = 0
elapsedTime = 0
args = build_argparser().parse_args()
#model_xml = "lrmodels/tiny-YoloV3/FP32/frozen_tiny_yolo_v3.xml" #<--- CPU
model_xml = "/home/saket/openvino_models/tiger/km_adam_yolov3_tinytiger_ckpt_9.xml" #<--- MYRIAD
model_bin = os.path.splitext(model_xml)[0] + ".bin"
'''
cap = cv2.VideoCapture(0)
cap.set(cv2.CAP_PROP_FPS, 30)
cap.set(cv2.CAP_PROP_FRAME_WIDTH, camera_width)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, camera_height)
'''
#cap = cv2.VideoCapture("data/input/testvideo.mp4")
#camera_width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
#camera_height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
#frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
#vidfps = int(cap.get(cv2.CAP_PROP_FPS))
#print("videosFrameCount =", str(frame_count))
#print("videosFPS =", str(vidfps))
time.sleep(1)
plugin = IEPlugin(device=args.device)
if "CPU" in args.device:
plugin.add_cpu_extension("lib/libcpu_extension.so")
net = IENetwork(model=model_xml, weights=model_bin)
input_blob = next(iter(net.inputs))
exec_net = plugin.load(network=net)
import glob
images = glob.glob('/media/saket/014178da-fdf2-462c-b901-d5f4dbce2e275/nn/PyTorch-YOLOv3/data/tigersamples/*.jpg')
print ('Num images',len(images))
counter = 0
while cv2.waitKey(0):
if cv2.waitKey(1)&0xFF == ord('q') or counter>len(images)-1:
break
t1 = time.time()
image = np.asarray(Image.open(images[counter]).convert('RGB'))
master = np.zeros((1920, 1920, 3),dtype='uint8')
master[420:1500,:]=image
image = master
camera_height, camera_width = image.shape[0:2]
new_w,new_h = 416,416
resized_image = cv2.resize(image, (new_w, new_h), interpolation = cv2.INTER_CUBIC)
canvas = np.full((m_input_size, m_input_size, 3), 0)
canvas[(m_input_size-new_h)//2:(m_input_size-new_h)//2 + new_h,(m_input_size-new_w)//2:(m_input_size-new_w)//2 + new_w, :] = resized_image
prepimg = canvas
prepimg = prepimg[np.newaxis, :, :, :] # Batch size axis add
prepimg = prepimg.transpose((0, 3, 1, 2)) # NHWC to NCHW
#print (prepimg.shape)
outputs = exec_net.infer(inputs={input_blob: prepimg})
objects = []
#output_name = detector/yolo-v3-tiny/Conv_12/BiasAdd/YoloRegion
#output_name = detector/yolo-v3-tiny/Conv_9/BiasAdd/YoloRegion
objects = []
for output in outputs.values():
objects = ParseYOLOV3Output(output,new_h, new_w, camera_height, camera_width, 0.5, objects)
# Filtering overlapping boxes
objlen = len(objects)
for i in range(objlen):
if (objects[i].confidence == 0.0):
continue
for j in range(i + 1, objlen):
if (IntersectionOverUnion(objects[i], objects[j]) >= 0.5):
if objects[i].confidence < objects[j].confidence:
objects[i], objects[j] = objects[j], objects[i]
objects[j].confidence = 0.0
# Drawing boxes
image = image[420:1500,:]
print (image.shape)
for obj in objects:
if obj.confidence < 0.5:
continue
label = obj.class_id
confidence = obj.confidence
if confidence > 0.5:
label_text = LABELS[label] + " (" + "{:.1f}".format(confidence * 100) + "%)"
cv2.rectangle(image, (obj.xmin, obj.ymin-420), (obj.xmax, obj.ymax-420), box_color, box_thickness)
cv2.putText(image, label_text, (obj.xmin, obj.ymin - 5), cv2.FONT_HERSHEY_SIMPLEX, 0.6, label_text_color, 1)
cv2.putText(image, fps, (camera_width - 170, 15), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (38, 0, 255), 1, cv2.LINE_AA)
cv2.imshow("Result", image)
elapsedTime = time.time() - t1
fps = "(Playback) {:.1f} FPS".format(1/elapsedTime)
## frame skip, video file only
#skip_frame = int((vidfps - int(1/elapsedTime)) / int(1/elapsedTime))
#framepos += skip_frame
counter= counter+1
cv2.destroyAllWindows()
del net
del exec_net
del plugin