def test_import():
cur_dir = os.path.dirname(os.path.abspath(__file__))
tflm_dir = os.path.abspath(cur_dir + '/../assets/tests')
tflm_name = 'mobilenet_v2_1.0_224.tflite'
path = os.path.join(tflm_dir, tflm_name)
with open(path, 'rb') as f:
buf = f.read()
model = Model.GetRootAsModel(buf, 0)
assert (model.Version() == 3)
def __init_op_info(self, model_path):
self.__tflite_ops = []
self.__tflite_op_types = []
data = open(model_path, "rb").read()
raw_model = Model.GetRootAsModel(bytearray(data), 0)
tflite_graph = raw_model.Subgraphs(0)
for idx in range(tflite_graph.OperatorsLength()):
op = tflite_graph.Operators(idx)
op_type = raw_model.OperatorCodes(op.OpcodeIndex()).BuiltinCode()
self.__tflite_ops.append(op)
self.__tflite_op_types.append(op_type)
def __analyze_tflite():
print('vistflite...')
with open(sys.argv[1], 'rb') as df:
content = df.read()
model = Model.GetRootAsModel(content, 0)
print(model.Version())
print(model.Description())
# OperatorCodes
for i in range(model.OperatorCodesLength()):
op = model.OperatorCodes(i)
if op.CustomCode():
print('Custom OP:', op.Version(), op.CustomCode())
else:
print('Builtin OP:', op.Version(), op.BuiltinCode(), getBuiltinOperatorStringName(op.BuiltinCode()))
# SubGraphs
print('SubGraphs:')
for i in range(model.SubgraphsLength()):
subGraph = model.Subgraphs(i)
print('Name:', subGraph.Name())
print('Tensors:')
for j in range(subGraph.TensorsLength()):
print(subGraph.Tensors(j).Name())
# we download model tar file and extract, finally get tflite file
model_path = download_testdata(model_url,
"mobilenet_v1_1.0_224_quant.tgz",
module=['tf', 'official'])
model_dir = os.path.dirname(model_path)
extract(model_path)
# now we have mobilenet_v1_1.0_224_quant.tflite on disk and open it
tflite_model_file = os.path.join(model_dir,
"mobilenet_v1_1.0_224_quant.tflite")
tflite_model_buf = open(tflite_model_file, "rb").read()
# get TFLite model from buffer
from tflite.Model import Model # edit here
tflite_model = Model.GetRootAsModel(tflite_model_buf, 0)
#######################################################################
# Generic run functions for TVM & TFLite
# --------------------------------------
target = tvm.target.riscv_cpu("spike")
input_tensor = "input"
input_shape = (1, 224, 224, 3)
input_dtype = "uint8"
# Parse TFLite model and convert it to a Relay module
mod, params = relay.frontend.from_tflite(
tflite_model,
shape_dict={input_tensor: input_shape},
dtype_dict={input_tensor: input_dtype})
""" tensorize flow
def read_tflite_model(path):
data = open(path, "rb").read()
model = Model.GetRootAsModel(bytearray(data), 0)
return model
def read_model(key):
with open(util_for_test.getPath(key), 'rb') as f:
buf = f.read()
model = Model.GetRootAsModel(buf, 0)
return model
def read_tflite_model(file):
buf = open(file, "rb").read()
buf = bytearray(buf)
model = Model.GetRootAsModel(buf, 0)
return model
def load_model(model_path):
# Load a model from a .tflite file at `model_path`
with open(model_path, 'rb') as f:
buf = f.read()
buf = bytearray(buf)
return Model.GetRootAsModel(buf, 0)
def test_import():
path = shrub.testing.download('mobilenet_v2_1.0_224.tflite')
with open(path, 'rb') as f:
buf = f.read()
model = Model.GetRootAsModel(buf, 0)
assert(model.Version() == 3)
