def test_optional_inputs(runner):
input_ids = np.array([1, 2]).astype(np.int32)
test_model = OptionalInputs()
exp0 = test_model(input_ids)
exp1 = test_model(input_ids, np.array([1, 2]).astype(np.int32))
oxml = keras2onnx.convert_keras(test_model)
assert runner('opt_inputs_0', oxml, [input_ids], exp0)
from onnxconverter_common.onnx_fx import GraphFunctionType as _Ty
oxml1 = keras2onnx.convert_keras(test_model, initial_types=(_Ty.I32(['N']), _Ty.I32(['N'])))
assert runner('opt_inputs_1', oxml1, [input_ids, np.array([1, 2]).astype(np.int32)], exp1)
apply_cast(scope, cast_batch, operator.output_full_names[2], container, to=onnx_proto.TensorProto.INT32)
apply_identity(scope, box_batch, operator.output_full_names[0], container)
apply_identity(scope, score_batch, operator.output_full_names[1], container)
set_converter(YOLONMSLayer, convert_NMSLayer)
yolo_model_graph_tiny = None
evaluation_model_graph_tiny = None
nms_model_graph_tiny = None
num_classes = 20
@Graph.trace(
input_types=[_Ty.F(shape=['N', 3, 'M1', 'M2']), _Ty.F(shape=['N', 2])],
output_types=[_Ty.F(shape=[1, 'M1', 4]), _Ty.F(shape=[1, num_classes, 'M2']), _Ty.I32(shape=[1, 'M3', 3])],
outputs=["yolonms_layer_1", "yolonms_layer_1_1", "yolonms_layer_1_2"])
def combine_model_tiny(input_1, image_shape):
global yolo_model_graph_tiny
global evaluation_model_graph_tiny
global nms_model_graph_tiny
output_1 = yolo_model_graph_tiny(input_1)
input_2 = output_1 + (image_shape,)
yolo_evaluation_layer_1, yolo_evaluation_layer_2 = evaluation_model_graph_tiny(*input_2)
nms_layer_1_1, nms_layer_1_2, nms_layer_1_3 = nms_model_graph_tiny(yolo_evaluation_layer_1, yolo_evaluation_layer_2)
return nms_layer_1_1, nms_layer_1_2, nms_layer_1_3
yolo_model_graph = None
evaluation_model_graph = None
nms_model_graph = None
set_converter(YOLONMSLayer, convert_NMSLayer)
yolo_model_graph_tiny = None
evaluation_model_graph_tiny = None
nms_model_graph_tiny = None
@Graph.trace(
input_types=[_Ty.F(shape=['N', 3, 'M1', 'M2']),
_Ty.F(shape=['N', 2])],
output_types=[
_Ty.F(shape=[1, 'M1', 4]),
_Ty.F(shape=[1, 80, 'M2']),
_Ty.I32(shape=[1, 'M3', 3])
],
outputs=["yolonms_layer_1", "yolonms_layer_1_1", "yolonms_layer_1_2"])
def combine_model_tiny(input_1, image_shape):
global yolo_model_graph_tiny
global evaluation_model_graph_tiny
global nms_model_graph_tiny
output_1 = yolo_model_graph_tiny(input_1)
input_2 = output_1 + (image_shape, )
yolo_evaluation_layer_1, yolo_evaluation_layer_2 = evaluation_model_graph_tiny(
*input_2)
nms_layer_1_1, nms_layer_1_2, nms_layer_1_3 = nms_model_graph_tiny(
yolo_evaluation_layer_1, yolo_evaluation_layer_2)
return nms_layer_1_1, nms_layer_1_2, nms_layer_1_3
cast_batch, op_version=operator.target_opset, axes=[0])
apply_cast(scope, cast_batch, operator.output_full_names[2], container, to=onnx_proto.TensorProto.INT32)
apply_identity(scope, box_batch, operator.output_full_names[0], container)
apply_identity(scope, score_batch, operator.output_full_names[1], container)
set_converter(YOLONMSLayer, convert_NMSLayer)
yolo_model_graph_tiny = None
evaluation_model_graph_tiny = None
nms_model_graph_tiny = None
@Graph.trace(
input_types=[_Ty.F(shape=['N', 3, 'M1', 'M2']), _Ty.F(shape=['N', 2])],
output_types=[_Ty.F(shape=[1, 'M1', 4]), _Ty.F(shape=[1, 80, 'M2']), _Ty.I32(shape=[1, 'M3', 3])],
outputs=["yolonms_layer_1", "yolonms_layer_1_1", "yolonms_layer_1_2"])
def combine_model_tiny(input_1, image_shape):
global yolo_model_graph_tiny
global evaluation_model_graph_tiny
global nms_model_graph_tiny
output_1 = yolo_model_graph_tiny(input_1)
input_2 = output_1 + (image_shape,)
yolo_evaluation_layer_1, yolo_evaluation_layer_2 = evaluation_model_graph_tiny(*input_2)
nms_layer_1_1, nms_layer_1_2, nms_layer_1_3 = nms_model_graph_tiny(yolo_evaluation_layer_1, yolo_evaluation_layer_2)
return nms_layer_1_1, nms_layer_1_2, nms_layer_1_3
yolo_model_graph = None
evaluation_model_graph = None
nms_model_graph = None