def init(config):
model = raimodel.Model.load(config['modelpath'])
host = config['server'].split(':')[0]
port = config['server'].split(':')[1]
init.con = rai.Client(host=host, port=port)
init.con.modelset('model', rai.Backend.torch, rai.Device.cpu, model)
image, init.img_class = get_one_image(transpose=(2, 0, 1))
init.image = rai.BlobTensor.from_numpy(image)
def init(config):
channel = grpc.insecure_channel('localhost:8500')
init.stub = prediction_service_pb2_grpc.PredictionServiceStub(channel)
init.request = predict_pb2.PredictRequest()
init.request.model_spec.name = 'resnet'
init.request.model_spec.signature_name = 'serving_default'
image, init.img_class = get_one_image()
init.image = tf.contrib.util.make_tensor_proto(image)
def init(config):
img, init.img_class = get_one_image(transpose=(2, 0, 1))
imgdata = ImageData()
# protobuf assumes the shape of the image is (1, 3, height, width)
# where 1 is the batchsize and 3 is number of channels
imgdata.image = img.tobytes()
imgdata.height = img.shape[2]
imgdata.width = img.shape[3]
imgdata.dtype = img.dtype.name
init.image = imgdata
channel = grpc.insecure_channel('localhost:50051')
init.stub = PredictorStub(channel)
def init(config):
with tf.gfile.GFile(config['modelpath'], "rb") as f:
restored_graph_def = tf.GraphDef()
restored_graph_def.ParseFromString(f.read())
with tf.Graph().as_default() as graph:
tf.import_graph_def(
restored_graph_def,
input_map=None,
return_elements=None,
name="")
init.graph = graph
init.image, init.img_class = get_one_image()
init.images_tensor = graph.get_tensor_by_name('images:0')
init.logits_tensor = graph.get_tensor_by_name('output:0')
def init(config):
host = config['server'].split(':')[0]
port = config['server'].split(':')[1]
init.con = rai.Client(host=host, port=port)
graph = raimodel.Model.load(config['modelpath'])
inputs = ['images']
outputs = ['output']
init.con.modelset('graph',
rai.Backend.tf,
rai.Device.cpu,
graph,
input=inputs,
output=outputs)
image, init.img_class = get_one_image()
init.image = rai.BlobTensor.from_numpy(image)
def init():
init.img_list, init.img_class = get_one_image(transpose=(2, 0, 1))
init.img_list = init.img_list.tolist()
def init():
init.img_list, init.img_class = get_one_image()
init.img_list = init.img_list.tolist()
def init(config):
init.model = torch.jit.load(config['modelpath'])
image, init.img_class = get_one_image(transpose=(2, 0, 1))
init.image = torch.from_numpy(image)