[MO MXNET] Fixed spatial reshape on GluonCV models (#587)
* added value propagation for slice_like op * Mark slice_lice as undead node * fixes in mark_undead_nodes and unittests update
This commit is contained in:
Yegor Kruglov authored and 
GitHub committed
1 parent
c9d4e6b934
commit
465707eba7
23 files changed
+426
-319
No files matched your search
@@ -178,6 +178,7 @@ extensions/front/mxnet/flatten_ext.py
|
||||
extensions/front/mxnet/fully_connected_ext.py
|
||||
extensions/front/mxnet/gather.py
|
||||
extensions/front/mxnet/gather_ext.py
|
||||
extensions/front/mxnet/gluoncv_ssd_anchors.py
|
||||
extensions/front/mxnet/instance_norm_ext.py
|
||||
extensions/front/mxnet/leaky_relu.py
|
||||
extensions/front/mxnet/lrn_ext.py
|
||||
@@ -481,7 +482,6 @@ extensions/middle/__init__.py
|
||||
extensions/middle/AddFakeQuantizeFuse.py
|
||||
extensions/middle/AddIsCyclicAttribute.py
|
||||
extensions/middle/AddMeanScaleValues.py
|
||||
extensions/middle/AnchorToPriorBox.py
|
||||
extensions/middle/ApplyNHWCtoNCHWpermutation.py
|
||||
extensions/middle/ApplyPermutations.py
|
||||
extensions/middle/ArgMaxToTopK.py
|
||||
@@ -556,7 +556,6 @@ extensions/middle/SliceConverter.py
|
||||
extensions/middle/SliceLikeToStridedSlice.py
|
||||
extensions/middle/sparse_reshape.py
|
||||
extensions/middle/SplitConcatPairToInterpolate.py
|
||||
extensions/middle/ssd_anchors_to_const.py
|
||||
extensions/middle/SwapAxesMiddleReplacer.py
|
||||
extensions/middle/TensorIterator_utils.py
|
||||
extensions/middle/TensorIteratorBackEdge.py
|
||||
|
||||
@@ -36,12 +36,12 @@ def graph_template(weights_initial_shape, new_reshape_shape, limits_initial_shap
|
||||
]
|
||||
|
||||
core_nodes = lambda weights_shape, limit_shape, reshape_shape: {
|
||||
**regular_op_with_shaped_data('input', None, {'type': 'Parameter'}),
|
||||
**regular_op_with_shaped_data('input', None, {'type': 'Parameter', 'op': 'Parameter'}),
|
||||
|
||||
**valued_const_with_data('weights', np.ones(weights_shape)),
|
||||
|
||||
**const_with_data('dim', int64_array(reshape_shape)),
|
||||
**regular_op_with_shaped_data('reshape', reshape_shape, {'type': 'Reshape', 'infer': Reshape.infer}),
|
||||
**regular_op_with_shaped_data('reshape', reshape_shape, {'type': 'Reshape', 'infer': Reshape.infer, 'op': 'Reshape'}),
|
||||
|
||||
**valued_const_with_data('il', np.ones(limit_shape)),
|
||||
**valued_const_with_data('ih', np.ones(limit_shape)),
|
||||
@@ -49,9 +49,9 @@ def graph_template(weights_initial_shape, new_reshape_shape, limits_initial_shap
|
||||
**valued_const_with_data('oh', np.ones(limit_shape)),
|
||||
|
||||
**regular_op_with_shaped_data('FQ', weights_shape, {'type': 'FakeQuantize', 'infer': FakeQuantize.infer,
|
||||
'stop_value_propagation': True, 'levels': 2}),
|
||||
'stop_value_propagation': True, 'levels': 2, 'op': 'FakeQuantize'}),
|
||||
|
||||
**regular_op_with_shaped_data('convolution', None, {'type': 'Convolution'}),
|
||||
**regular_op_with_shaped_data('convolution', None, {'type': 'Convolution', 'op': 'Convolution'}),
|
||||
|
||||
**result(),
|
||||
}
|
||||
|
||||
@@ -31,24 +31,24 @@ nodes_attributes = {
|
||||
'placeholder_1_data': {'name': 'placeholder_1_data', 'value': None, 'shape': None, 'kind': 'data',
|
||||
'data_type': None},
|
||||
# Transpose layers
|
||||
'const_1': {'value': None, 'kind': 'op', 'type': 'Const'},
|
||||
'const_1': {'value': None, 'kind': 'op', 'type': 'Const', 'op': 'Const'},
|
||||
'const_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
|
||||
'permute_1': {'type': 'Transpose', 'value': None, 'kind': 'op', 'op': 'Transpose'},
|
||||
'permute_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
|
||||
'const_2': {'value': None, 'kind': 'op', 'type': 'Const'},
|
||||
'const_2': {'value': None, 'kind': 'op', 'type': 'Const', 'op': 'Const'},
|
||||
'const_2_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
|
||||
'permute_2': {'type': 'Transpose', 'value': None, 'kind': 'op', 'op': 'Transpose'},
|
||||
'permute_2_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
|
||||
'const_3': {'value': None, 'kind': 'op', 'type': 'Const'},
|
||||
'const_3': {'value': None, 'kind': 'op', 'type': 'Const', 'op': 'Const'},
|
||||
'const_3_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
|
||||
'permute_3': {'type': 'Transpose', 'value': None, 'kind': 'op', 'op': 'Transpose'},
|
||||
'permute_3_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'op_output': { 'op': 'Result', 'kind': 'op'}
|
||||
'op_output': {'op': 'Result', 'kind': 'op'}
|
||||
}
|
||||
|
||||
|
||||
|
||||
@@ -25,21 +25,22 @@ from mo.utils.unittest.graph import build_graph, result, regular_op_with_shaped_
|
||||
connect_data
|
||||
|
||||
nodes = {
|
||||
**regular_op_with_shaped_data('placeholder', [1, 3, 30, 40], {'type': 'Parameter'}),
|
||||
**regular_op_with_shaped_data('placeholder', [1, 3, 30, 40], {'type': 'Parameter', 'op': 'Parameter'}),
|
||||
**valued_const_with_data('out_shape', np.array([60, 160])),
|
||||
|
||||
**regular_op_with_shaped_data('interpolate', [1, 3, 60, 160], {'type': 'Interpolate', 'axes': [2, 3]}),
|
||||
**regular_op_with_shaped_data('interpolate', [1, 3, 60, 160], {'type': 'Interpolate', 'axes': [2, 3],
|
||||
'op': 'Interpolate'}),
|
||||
|
||||
**regular_op_with_shaped_data('shape', [4], {'type': 'ShapeOf'}),
|
||||
**regular_op_with_shaped_data('shape', [4], {'type': 'ShapeOf', 'op': 'ShapeOf'}),
|
||||
**valued_const_with_data('indices', np.array([2, 3])),
|
||||
**valued_const_with_data('axis', np.array(0)),
|
||||
**regular_op_with_shaped_data('gather', [2], {'type': 'Gather'}),
|
||||
**regular_op_with_shaped_data('gather', [2], {'type': 'Gather', 'op': 'Gather'}),
|
||||
|
||||
**valued_const_with_data('multiplier', np.array([2, 4])),
|
||||
**regular_op_with_shaped_data('mul', [2], {'type': 'Multiply'}),
|
||||
**regular_op_with_shaped_data('mul', [2], {'type': 'Multiply', 'op': 'Mul'}),
|
||||
|
||||
**regular_op_with_shaped_data('placeholder_1', [1, 3, 60, 160], {'type': 'Parameter'}),
|
||||
**regular_op_with_shaped_data('concat', [1, 7, 60, 160], {'type': 'Concat', 'axis': 1}),
|
||||
**regular_op_with_shaped_data('placeholder_1', [1, 3, 60, 160], {'type': 'Parameter', 'op': 'Parameter'}),
|
||||
**regular_op_with_shaped_data('concat', [1, 7, 60, 160], {'type': 'Concat', 'axis': 1, 'op': 'Concat'}),
|
||||
|
||||
**result(),
|
||||
}
|
||||
|
||||
@@ -34,36 +34,36 @@ nodes_attributes = {
|
||||
'placeholder_data': {'kind': 'data'},
|
||||
|
||||
# weights
|
||||
'weights_const': {'type': 'Const', 'kind': 'op', 'value': np.array([], dtype=np.float32)},
|
||||
'weights_const': {'type': 'Const', 'kind': 'op', 'value': np.array([], dtype=np.float32), 'op': 'Const'},
|
||||
'weights_data': {'kind': 'data'},
|
||||
|
||||
# quantize
|
||||
'quantize1': {'type': 'FakeQuantize', 'kind': 'op', 'levels': 5},
|
||||
'quantize2': {'type': 'FakeQuantize', 'kind': 'op', 'levels': 2},
|
||||
'quantize3': {'type': 'FakeQuantize', 'kind': 'op', 'levels': None},
|
||||
'quantize4': {'type': 'FakeQuantize', 'kind': 'op', 'levels': 122},
|
||||
'quantize5': {'type': 'FakeQuantize', 'kind': 'op', 'levels': 202},
|
||||
'quantize6': {'type': 'FakeQuantize', 'kind': 'op', 'levels': 257},
|
||||
'quantize1': {'type': 'FakeQuantize', 'kind': 'op', 'levels': 5, 'op': 'FakeQuantize'},
|
||||
'quantize2': {'type': 'FakeQuantize', 'kind': 'op', 'levels': 2, 'op': 'FakeQuantize'},
|
||||
'quantize3': {'type': 'FakeQuantize', 'kind': 'op', 'levels': None, 'op': 'FakeQuantize'},
|
||||
'quantize4': {'type': 'FakeQuantize', 'kind': 'op', 'levels': 122, 'op': 'FakeQuantize'},
|
||||
'quantize5': {'type': 'FakeQuantize', 'kind': 'op', 'levels': 202, 'op': 'FakeQuantize'},
|
||||
'quantize6': {'type': 'FakeQuantize', 'kind': 'op', 'levels': 257, 'op': 'FakeQuantize'},
|
||||
'quantize_data': {'kind': 'data'},
|
||||
'new_quantize1': {'kind': 'op', 'type': 'FakeQuantize', 'levels': 5},
|
||||
'new_quantize4': {'kind': 'op', 'type': 'FakeQuantize', 'levels': 122},
|
||||
'new_quantize5': {'kind': 'op', 'type': 'FakeQuantize', 'levels': 202},
|
||||
'new_quantize1': {'kind': 'op', 'type': 'FakeQuantize', 'levels': 5, 'op': 'FakeQuantize'},
|
||||
'new_quantize4': {'kind': 'op', 'type': 'FakeQuantize', 'levels': 122, 'op': 'FakeQuantize'},
|
||||
'new_quantize5': {'kind': 'op', 'type': 'FakeQuantize', 'levels': 202, 'op': 'FakeQuantize'},
|
||||
'new_quantize_data': {'kind': 'data'},
|
||||
|
||||
# quantize input/output
|
||||
'output_high_init': {'kind': 'op', 'type': 'Const'},
|
||||
'output_high_init': {'kind': 'op', 'type': 'Const', 'op': 'Const'},
|
||||
'output_high_init_data': {'kind': 'data', 'value': 3},
|
||||
'output_low_init': {'kind': 'op', 'type': 'Const'},
|
||||
'output_low_init': {'kind': 'op', 'type': 'Const', 'op': 'Const'},
|
||||
'output_low_init_data': {'kind': 'data', 'value': -1.5},
|
||||
|
||||
'input_low': {'kind': 'op', 'type': 'Const'},
|
||||
'input_low': {'kind': 'op', 'type': 'Const', 'op': 'Const'},
|
||||
'input_low_data': {'kind': 'data'},
|
||||
'input_high': {'kind': 'op', 'type': 'Const'},
|
||||
'input_high': {'kind': 'op', 'type': 'Const', 'op': 'Const'},
|
||||
'input_high_data': {'kind': 'data'},
|
||||
|
||||
'output_low': {'kind': 'op', 'type': 'Const'},
|
||||
'output_low': {'kind': 'op', 'type': 'Const', 'op': 'Const'},
|
||||
'output_low_data': {'kind': 'data'},
|
||||
'output_high': {'kind': 'op', 'type': 'Const'},
|
||||
'output_high': {'kind': 'op', 'type': 'Const', 'op': 'Const'},
|
||||
'output_high_data': {'kind': 'data'},
|
||||
|
||||
'output_high_init_data1': {'kind': 'data', 'value': 256.1},
|
||||
@@ -76,58 +76,58 @@ nodes_attributes = {
|
||||
'mul': {'kind': 'op', 'op': 'Mul'},
|
||||
'add': {'kind': 'op', 'op': 'Add'},
|
||||
|
||||
'scale': {'kind': 'op', 'type': 'Const', 'value': 1.125},
|
||||
'shift': {'kind': 'op', 'type': 'Const', 'value': -1.5},
|
||||
'scale1': {'kind': 'op', 'type': 'Const', 'value': 1.9735537190082646},
|
||||
'shift1': {'kind': 'op', 'type': 'Const', 'value': 17.3},
|
||||
'scale2': {'kind': 'op', 'type': 'Const', 'value': 0.010711442786069652},
|
||||
'shift2': {'kind': 'op', 'type': 'Const', 'value': -2.573},
|
||||
'scale': {'kind': 'op', 'type': 'Const', 'value': 1.125, 'op': 'Const'},
|
||||
'shift': {'kind': 'op', 'type': 'Const', 'value': -1.5, 'op': 'Const'},
|
||||
'scale1': {'kind': 'op', 'type': 'Const', 'value': 1.9735537190082646, 'op': 'Const'},
|
||||
'shift1': {'kind': 'op', 'type': 'Const', 'value': 17.3, 'op': 'Const'},
|
||||
'scale2': {'kind': 'op', 'type': 'Const', 'value': 0.010711442786069652, 'op': 'Const'},
|
||||
'shift2': {'kind': 'op', 'type': 'Const', 'value': -2.573, 'op': 'Const'},
|
||||
|
||||
'shift_data': {'kind': 'data'},
|
||||
'scale_data': {'kind': 'data'},
|
||||
'mul_data': {'kind': 'data'},
|
||||
'add_data': {'kind': 'data'},
|
||||
|
||||
'convolution': {'type': 'Convolution', 'kind': 'op'},
|
||||
'convert': {'type': 'Convert', 'kind': 'op', 'dst_type': np.float32},
|
||||
'convolution': {'type': 'Convolution', 'kind': 'op', 'op': 'Convolution'},
|
||||
'convert': {'type': 'Convert', 'kind': 'op', 'dst_type': np.float32, 'op': 'Cast'},
|
||||
'convert_data': {'kind': 'data'},
|
||||
|
||||
'result_data': {'kind':'data'},
|
||||
'result_data': {'kind': 'data'},
|
||||
'result': {'kind': 'op', 'op': 'Result'},
|
||||
|
||||
# accuracy test
|
||||
'ac_weights': {'kind': 'op', 'op': 'Const', 'shape': None, 'value': None, 'infer': Const.infer},
|
||||
'ac_weights_data': {'kind': 'data', 'shape': None, 'value': None},
|
||||
|
||||
'ac_input_low': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer},
|
||||
'ac_input_low': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer, 'op': 'Const'},
|
||||
'ac_input_low_data': {'kind': 'data', 'value': None, 'shape': None},
|
||||
'ac_input_high': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer},
|
||||
'ac_input_high': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer, 'op': 'Const'},
|
||||
'ac_input_high_data': {'kind': 'data', 'value': None, 'shape': None},
|
||||
'ac_output_low': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer},
|
||||
'ac_output_low': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer, 'op': 'Const'},
|
||||
'ac_output_low_data': {'kind': 'data', 'value': None, 'shape': None},
|
||||
'ac_output_high': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer},
|
||||
'ac_output_high': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer, 'op': 'Const'},
|
||||
'ac_output_high_data': {'kind': 'data', 'value': None, 'shape': None},
|
||||
|
||||
'ac_fakeQuantize': {'kind': 'op', 'type': 'FakeQuantize', 'levels': None, 'infer': FakeQuantize.infer},
|
||||
'ac_fakeQuantize': {'kind': 'op', 'type': 'FakeQuantize', 'levels': None, 'infer': FakeQuantize.infer, 'op': 'FakeQuantize'},
|
||||
'ac_fakeQuantize_data': {'kind': 'data', 'shape': None, 'value': None},
|
||||
'ac_quantize': {'kind': 'op', 'type': 'fakeQuantize', 'levels': None, 'infer': FakeQuantize.infer},
|
||||
'ac_quantize': {'kind': 'op', 'type': 'fakeQuantize', 'levels': None, 'infer': FakeQuantize.infer, 'op': 'FakeQuantize'},
|
||||
'ac_quantize_data': {'kind': 'data', 'shape': None, 'value': None},
|
||||
|
||||
'ac_convolution': {'kind': 'op', 'type': 'Convolution'},
|
||||
'ac_convolution': {'kind': 'op', 'type': 'Convolution', 'op': 'Convolution'},
|
||||
|
||||
'ac_mul': {'kind': 'op', 'op': 'Mul', 'infer': lambda node: eltwise_infer(node, lambda a, b: a * b)},
|
||||
'ac_mul_data': {'kind': 'data', 'shape': None, 'value': None},
|
||||
'ac_add': {'kind': 'op', 'op': 'Add', 'infer': lambda node: eltwise_infer(node, lambda a, b: a + b)},
|
||||
'ac_add_data': {'kind': 'data', 'shape': None, 'value': None},
|
||||
|
||||
'ac_scale': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer},
|
||||
'ac_scale': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer, 'op': 'Const'},
|
||||
'ac_scale_data': {'kind': 'data', 'shape': None, 'value': None},
|
||||
'ac_shift': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer},
|
||||
'ac_shift': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer, 'op': 'Const'},
|
||||
'ac_shift_data': {'kind': 'data', 'shape': None, 'value': None},
|
||||
|
||||
'ac_output_low_ref': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer},
|
||||
'ac_output_low_ref': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer, 'op': 'Const'},
|
||||
'ac_output_low_ref_data': {'kind': 'data', 'shape': None, 'value': None},
|
||||
'ac_output_high_ref': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer},
|
||||
'ac_output_high_ref': {'kind': 'op', 'type': 'Const', 'shape': None, 'value': None, 'infer': Const.infer, 'op': 'Const'},
|
||||
'ac_output_high_ref_data': {'kind': 'data', 'shape': None, 'value': None}
|
||||
}
|
||||
|
||||
@@ -838,7 +838,8 @@ class CompressionDataTypeTest(unittest.TestCase):
|
||||
**valued_const_with_data('weights', np.ones([1, 2, 3, 4], dtype=original)),
|
||||
|
||||
**valued_const_with_data('int_weights', np.ones([1, 2, 3, 4], dtype=np.uint8)),
|
||||
**regular_op_with_shaped_data('cast', [1, 2, 3, 4], {'type': 'Convert', 'dst_type': transformed}),
|
||||
**regular_op_with_shaped_data('cast', [1, 2, 3, 4], {'type': 'Convert', 'dst_type': transformed,
|
||||
'op': 'Cast'}),
|
||||
|
||||
**valued_const_with_data('il', np.array([0])),
|
||||
**valued_const_with_data('ih', np.array([254])),
|
||||
@@ -846,7 +847,8 @@ class CompressionDataTypeTest(unittest.TestCase):
|
||||
**valued_const_with_data('oh', np.array([254])),
|
||||
|
||||
**regular_op_with_shaped_data('FQ', [1, 2, 3, 4], {'type': 'FakeQuantize', 'infer': FakeQuantize.infer,
|
||||
'stop_value_propagation': True, 'levels': 255}),
|
||||
'stop_value_propagation': True, 'levels': 255,
|
||||
'op': 'FakeQuantize'}),
|
||||
**result()
|
||||
}
|
||||
|
||||
|
||||
@@ -26,23 +26,23 @@ from mo.utils.unittest.graph import build_graph, result, regular_op_with_shaped_
|
||||
connect_data
|
||||
|
||||
nodes = {
|
||||
**regular_op_with_shaped_data('placeholder', [1, 3, 30, 40], {'type': 'Parameter'}),
|
||||
**regular_op_with_shaped_data('placeholder', [1, 3, 30, 40], {'type': 'Parameter', 'op': 'Parameter'}),
|
||||
**valued_const_with_data('out_shape', np.array([60, 160])),
|
||||
|
||||
**regular_op_with_shaped_data('interpolate', [1, 3, 60, 160],
|
||||
{'type': 'Interpolate', 'axes': int64_array([2, 3])}),
|
||||
**regular_op_with_shaped_data('identity_00', [1, 3, 60, 160], {'identity': True}),
|
||||
**regular_op_with_shaped_data('identity_01', [1, 3, 60, 160], {'identity': True}),
|
||||
{'type': 'Interpolate', 'axes': int64_array([2, 3]), 'op': 'Interpolate'}),
|
||||
**regular_op_with_shaped_data('identity_00', [1, 3, 60, 160], {'identity': True, 'op': 'Identity'}),
|
||||
**regular_op_with_shaped_data('identity_01', [1, 3, 60, 160], {'identity': True, 'op': 'Identity'}),
|
||||
|
||||
**regular_op_with_shaped_data('shape', [4], {'type': 'ShapeOf'}),
|
||||
**regular_op_with_shaped_data('shape', [4], {'type': 'ShapeOf', 'op': 'ShapeOf'}),
|
||||
**valued_const_with_data('indices', np.array([2, 3])),
|
||||
**valued_const_with_data('axis', np.array(0)),
|
||||
**regular_op_with_shaped_data('gather', [2], {'type': 'Gather'}),
|
||||
**regular_op_with_shaped_data('gather', [2], {'type': 'Gather', 'op': 'Gather'}),
|
||||
|
||||
**regular_op_with_shaped_data('placeholder_1', [1, 3, 60, 160], {'type': 'Parameter'}),
|
||||
**regular_op_with_shaped_data('identity_10', [1, 3, 60, 160], {'identity': True}),
|
||||
**regular_op_with_shaped_data('identity_11', [1, 3, 60, 160], {'identity': True}),
|
||||
**regular_op_with_shaped_data('concat', [1, 7, 60, 160], {'type': 'Concat', 'axis': 1}),
|
||||
**regular_op_with_shaped_data('placeholder_1', [1, 3, 60, 160], {'type': 'Parameter', 'op': 'Parameter'}),
|
||||
**regular_op_with_shaped_data('identity_10', [1, 3, 60, 160], {'identity': True, 'op': 'Identity'}),
|
||||
**regular_op_with_shaped_data('identity_11', [1, 3, 60, 160], {'identity': True, 'op': 'Identity'}),
|
||||
**regular_op_with_shaped_data('concat', [1, 7, 60, 160], {'type': 'Concat', 'axis': 1, 'op': 'Concat'}),
|
||||
|
||||
**result(),
|
||||
}
|
||||
|
||||
@@ -0,0 +1,126 @@
|
||||
"""
|
||||
Copyright (C) 2018-2020 Intel Corporation
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
"""
|
||||
import numpy as np
|
||||
|
||||
from typing import Dict
|
||||
|
||||
from extensions.front.mxnet.mx_reshape_to_reshape import MXReshapeToReshape
|
||||
from extensions.front.mxnet.ssd_detection_output_replacer import SsdPatternDetectionOutputReplacer
|
||||
from extensions.ops.elementwise import Div, Add, Sub
|
||||
from extensions.ops.split import Split
|
||||
from mo.front.common.partial_infer.utils import int64_array
|
||||
from mo.front.common.replacement import FrontReplacementPattern
|
||||
from mo.front.tf.graph_utils import create_op_node_with_second_input
|
||||
from mo.graph.graph import Graph, Node
|
||||
from mo.graph.port import Port
|
||||
from mo.middle.passes.convert_data_type import data_type_str_to_np
|
||||
from mo.ops.concat import Concat
|
||||
from mo.ops.reshape import Reshape
|
||||
|
||||
|
||||
def calculate_prior_box_value(value: Node, value_to_div: Port, value_to_add: Port):
|
||||
"""
|
||||
:param value: Node with value. Here is supposed the node with op='Split'
|
||||
:param value_to_div: Output port with values to be divided by 2
|
||||
:param value_to_add: Output port with values to be added to values from value_to_div port
|
||||
:return: Sub and Add nodes
|
||||
|
||||
The sub-graph can be described by formulas:
|
||||
min = value[value_to_add] - (value[value_to_div] / 2)
|
||||
max = value[value_to_add] + (value[value_to_div] / 2)
|
||||
"""
|
||||
graph = value.graph
|
||||
dtype = data_type_str_to_np(graph.graph['cmd_params'].data_type)
|
||||
_min = Sub(graph, dict(name=value.name + '/Sub')).create_node()
|
||||
div = create_op_node_with_second_input(graph, Div, np.array([2], dtype=dtype), op_attrs=dict(name=value.name + '/Div'))
|
||||
div.in_port(0).connect(value_to_div)
|
||||
_min.in_port(0).connect(value_to_add)
|
||||
_min.in_port(1).connect(div.out_port(0))
|
||||
|
||||
_max = Add(graph, dict(name=value.name + '/Add')).create_node()
|
||||
_max.in_port(0).connect(div.out_port(0))
|
||||
_max.in_port(1).connect(value_to_add)
|
||||
|
||||
return _min, _max
|
||||
|
||||
|
||||
class SsdAnchorsReplacer(FrontReplacementPattern):
|
||||
"""
|
||||
Replacing sub-graph with all anchors to sub-graph which calculates prior boxes values by formulas:
|
||||
|
||||
value[i] = xmin = value[i] - (value[i + 2] / 2)
|
||||
value[i + 1] = ymin = value[i + 1] - (value[i + 3] / 2)
|
||||
value[i + 2] = xmax = value[i] + (value[i + 2] / 2)
|
||||
value[i + 3] = ymax = value[i + 1] + (value[i + 3] / 2)
|
||||
"""
|
||||
|
||||
enabled = True
|
||||
graph_condition = [lambda graph: graph.graph['cmd_params'].enable_ssd_gluoncv]
|
||||
|
||||
def run_after(self):
|
||||
return [SsdPatternDetectionOutputReplacer, MXReshapeToReshape]
|
||||
|
||||
def pattern(self):
|
||||
return dict(
|
||||
nodes=[
|
||||
('slice_like', dict(op='slice_like')),
|
||||
('reshape0', dict(op='Reshape')),
|
||||
('reshape1', dict(op='Reshape')),
|
||||
('reshape2', dict(op='Reshape')),
|
||||
('reshape3', dict(op='Reshape')),
|
||||
('concat', dict(op='Concat')),
|
||||
('detection_output', dict(op='DetectionOutput'))
|
||||
],
|
||||
edges=[
|
||||
('slice_like', 'reshape0'),
|
||||
('reshape0', 'reshape1'),
|
||||
('reshape1', 'reshape2'),
|
||||
('reshape2', 'reshape3'),
|
||||
('reshape3', 'concat', {'in': 0}),
|
||||
('concat', 'detection_output', {'in': 2})
|
||||
])
|
||||
|
||||
def replace_pattern(self, graph: Graph, match: Dict[str, Node]):
|
||||
concat_node = match['concat']
|
||||
concat_node['axis'] = 1
|
||||
concat_name = concat_node.soft_get('name', concat_node.id)
|
||||
|
||||
concat_reshape = create_op_node_with_second_input(graph, Reshape, int64_array([1, 2, -1]), op_attrs=dict(
|
||||
name=concat_name + '/Reshape'))
|
||||
split_node = create_op_node_with_second_input(graph, Split, int64_array(1), op_attrs=dict(
|
||||
name=concat_name + '/Split', num_splits=2), input_node=concat_reshape)
|
||||
split_node_reshape = create_op_node_with_second_input(graph, Reshape, int64_array([-1, 4]), op_attrs=dict(
|
||||
name=split_node.name + '/Reshape'))
|
||||
split_node.out_port(0).connect(split_node_reshape.in_port(0))
|
||||
value = create_op_node_with_second_input(graph, Split, int64_array(1), op_attrs=dict(
|
||||
name=split_node_reshape.name + '/Split', num_splits=4), input_node=split_node_reshape)
|
||||
|
||||
xmin, xmax = calculate_prior_box_value(value, value_to_div=value.out_port(2), value_to_add=value.out_port(0))
|
||||
ymin, ymax = calculate_prior_box_value(value, value_to_div=value.out_port(3), value_to_add=value.out_port(1))
|
||||
|
||||
concat_slice_value = Concat(graph, dict(name=value.name + '/Concat', in_ports_count=4, axis=1)).create_node()
|
||||
for ind, node in enumerate([xmin, ymin, xmax, ymax]):
|
||||
concat_slice_value.in_port(ind).connect(node.out_port(0))
|
||||
|
||||
reshape_concat_values = create_op_node_with_second_input(graph, Reshape, int64_array([1, 1, -1]),
|
||||
op_attrs=dict(name=concat_slice_value.name + '/Reshape'),
|
||||
input_node=concat_slice_value)
|
||||
concat = Concat(graph, dict(name=reshape_concat_values.name + '/Concat', in_ports_count=2, axis=1)).create_node()
|
||||
concat.in_port(0).connect(reshape_concat_values.out_port(0))
|
||||
concat.in_port(1).connect(split_node.out_port(1))
|
||||
|
||||
match['detection_output'].in_port(2).get_connection().set_source(concat.out_port(0))
|
||||
concat_node.out_port(0).get_connection().set_destination(concat_reshape.in_port(0))
|
||||
@@ -0,0 +1,130 @@
|
||||
"""
|
||||
Copyright (C) 2018-2020 Intel Corporation
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
"""
|
||||
import numpy as np
|
||||
import unittest
|
||||
|
||||
from extensions.front.mxnet.gluoncv_ssd_anchors import SsdAnchorsReplacer
|
||||
from mo.front.common.partial_infer.utils import int64_array
|
||||
from mo.utils.ir_engine.compare_graphs import compare_graphs
|
||||
from mo.utils.unittest.graph import build_graph
|
||||
|
||||
nodes_attributes = {
|
||||
'slice_like': {'kind': 'op', 'op': 'slice_like'},
|
||||
'model_reshape0': {'kind': 'op', 'op': 'Reshape'},
|
||||
'model_reshape0_const': {'kind': 'op', 'op': 'Const', 'value': int64_array([1, -1, 4])},
|
||||
'model_reshape1': {'kind': 'op', 'op': 'Reshape'},
|
||||
'model_reshape1_const': {'kind': 'op', 'op': 'Const', 'value': int64_array([1, -1, 4])},
|
||||
'model_reshape2': {'kind': 'op', 'op': 'Reshape'},
|
||||
'model_reshape2_const': {'kind': 'op', 'op': 'Const', 'value': int64_array([1, -1])},
|
||||
'reshape0': {'kind': 'op', 'op': 'Reshape'},
|
||||
'reshape0_const': {'kind': 'op', 'op': 'Const', 'value': int64_array([1, -1])},
|
||||
'concat': {'kind': 'op', 'op': 'Concat'},
|
||||
'reshape1': {'kind': 'op', 'op': 'Reshape'},
|
||||
'reshape1_const': {'kind': 'op', 'op': 'Const', 'value': int64_array([1, 2, -1])},
|
||||
'split': {'kind': 'op', 'op': 'Split', 'num_splits': 2},
|
||||
'split_const': {'kind': 'op', 'op': 'Const', 'value': int64_array(1)},
|
||||
'reshape2': {'kind': 'op', 'op': 'Reshape'},
|
||||
'reshape2_const': {'kind': 'op', 'op': 'Const', 'value': int64_array([-1, 4])},
|
||||
'value': {'kind': 'op', 'op': 'Split', 'num_splits': 4},
|
||||
'value_const': {'kind': 'op', 'op': 'Const', 'value': int64_array(1)},
|
||||
'div_1': {'kind': 'op', 'op': 'Div'},
|
||||
'div_1_const': {'kind': 'op', 'op': 'Const', 'value': np.array([2], dtype=np.float32)},
|
||||
'div_2': {'kind': 'op', 'op': 'Div'},
|
||||
'div_2_const': {'kind': 'op', 'op': 'Const', 'value': np.array([2], dtype=np.float32)},
|
||||
'xmin': {'kind': 'op', 'op': 'Sub'},
|
||||
'ymin': {'kind': 'op', 'op': 'Sub'},
|
||||
'xmax': {'kind': 'op', 'op': 'Add'},
|
||||
'ymax': {'kind': 'op', 'op': 'Add'},
|
||||
'concat_value': {'kind': 'op', 'op': 'Concat', 'axis': 1},
|
||||
'reshape3': {'kind': 'op', 'op': 'Reshape'},
|
||||
'reshape3_const': {'kind': 'op', 'op': 'Const', 'value': int64_array([1, 1, -1])},
|
||||
'end_concat': {'kind': 'op', 'op': 'Concat'},
|
||||
'detection_output': {'kind': 'op', 'op': 'DetectionOutput'}
|
||||
}
|
||||
|
||||
|
||||
class SsdAnchorsReplacerTest(unittest.TestCase):
|
||||
|
||||
def test_replacer(self):
|
||||
graph = build_graph(
|
||||
nodes_attrs=nodes_attributes,
|
||||
edges=[
|
||||
('slice_like', 'model_reshape0', {'in': 0}),
|
||||
('model_reshape0_const', 'model_reshape0', {'in': 1}),
|
||||
('model_reshape0', 'model_reshape1', {'in': 0}),
|
||||
('model_reshape1_const', 'model_reshape1', {'in': 1}),
|
||||
('model_reshape1', 'model_reshape2', {'in': 0}),
|
||||
('model_reshape2_const', 'model_reshape2', {'in': 1}),
|
||||
('model_reshape2', 'reshape0', {'in': 0}),
|
||||
('reshape0_const', 'reshape0', {'in': 1}),
|
||||
('reshape0', 'concat'),
|
||||
('concat', 'detection_output', {'in': 2})
|
||||
],
|
||||
nodes_with_edges_only=True
|
||||
)
|
||||
|
||||
ref_graph = build_graph(
|
||||
nodes_attrs=nodes_attributes,
|
||||
edges=[
|
||||
('slice_like', 'model_reshape0', {'in': 0}),
|
||||
('model_reshape0_const', 'model_reshape0', {'in': 1}),
|
||||
('model_reshape0', 'model_reshape1', {'in': 0}),
|
||||
('model_reshape1_const', 'model_reshape1', {'in': 1}),
|
||||
('model_reshape1', 'model_reshape2', {'in': 0}),
|
||||
('model_reshape2_const', 'model_reshape2', {'in': 1}),
|
||||
('model_reshape2', 'reshape0', {'in': 0}),
|
||||
('reshape0_const', 'reshape0', {'in': 1}),
|
||||
('reshape0', 'concat'),
|
||||
('concat', 'reshape1', {'in': 0}),
|
||||
('reshape1_const', 'reshape1', {'in': 1}),
|
||||
('reshape1', 'split', {'in': 0}),
|
||||
('split_const', 'split', {'in': 1}),
|
||||
('split', 'reshape2', {'out': 0, 'in': 0}),
|
||||
('reshape2_const', 'reshape2', {'in': 1}),
|
||||
('reshape2', 'value', {'in': 0}),
|
||||
('value_const', 'value', {'in': 1}),
|
||||
('value', 'xmin', {'out': 0, 'in': 0}),
|
||||
('value', 'ymin', {'out': 1, 'in': 0}),
|
||||
('value', 'xmax', {'out': 0, 'in': 1}),
|
||||
('value', 'ymax', {'out': 1, 'in': 1}),
|
||||
('value', 'div_1', {'out': 2, 'in': 0}),
|
||||
('value', 'div_2', {'out': 3, 'in': 0}),
|
||||
('div_1_const', 'div_1', {'in': 1}),
|
||||
('div_2_const', 'div_2', {'in': 1}),
|
||||
('div_1', 'xmin', {'in': 1, 'out': 0}),
|
||||
('div_1', 'xmax', {'in': 0, 'out': 0}),
|
||||
('div_2', 'ymin', {'in': 1, 'out': 0}),
|
||||
('div_2', 'ymax', {'in': 0, 'out': 0}),
|
||||
('xmin', 'concat_value', {'in': 0}),
|
||||
('ymin', 'concat_value', {'in': 1}),
|
||||
('xmax', 'concat_value', {'in': 2}),
|
||||
('ymax', 'concat_value', {'in': 3}),
|
||||
('concat_value', 'reshape3', {'in': 0}),
|
||||
('reshape3_const', 'reshape3', {'in': 1}),
|
||||
('reshape3', 'end_concat', {'in': 0}),
|
||||
('split', 'end_concat', {'in': 1}),
|
||||
('end_concat', 'detection_output', {'in': 2})
|
||||
],
|
||||
update_attributes={
|
||||
'concat': {'axis': 1}
|
||||
},
|
||||
nodes_with_edges_only=True
|
||||
)
|
||||
graph.stage = 'front'
|
||||
graph.graph['cmd_params'].data_type = 'FP32'
|
||||
SsdAnchorsReplacer().find_and_replace_pattern(graph)
|
||||
flag, resp = compare_graphs(graph, ref_graph, 'detection_output', check_op_attrs=True)
|
||||
self.assertTrue(flag, resp)
|
||||
@@ -24,7 +24,6 @@ from mo.front.common.replacement import FrontReplacementSubgraph
|
||||
from mo.graph.graph import Graph, Node
|
||||
from mo.middle.pattern_match import find_pattern_matches
|
||||
from mo.ops.const import Const
|
||||
from mo.ops.crop import Crop
|
||||
|
||||
|
||||
class SsdPatternAnchorReshape(FrontReplacementSubgraph):
|
||||
|
||||
@@ -103,7 +103,7 @@ class SsdPatternDetectionOutputReplacer(FrontReplacementSubgraph):
|
||||
|
||||
def reshape_priorboxes(self, concat):
|
||||
for i, node in concat.in_nodes().items():
|
||||
reshape_node = create_op_node_with_second_input(concat.graph, Reshape, int64_array([0, 2, -1]),
|
||||
reshape_node = create_op_node_with_second_input(concat.graph, Reshape, int64_array([1, -1]),
|
||||
dict(name=concat.name + str(i) + '/PriorBoxReshape_'))
|
||||
node.out_port(0).disconnect()
|
||||
node.out_port(0).connect(reshape_node.in_port(0))
|
||||
|
||||
@@ -1,71 +0,0 @@
|
||||
"""
|
||||
Copyright (C) 2018-2020 Intel Corporation
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
"""
|
||||
|
||||
import numpy as np
|
||||
|
||||
from extensions.middle.SliceLikeToStridedSlice import SliceLikeToStridedSlice
|
||||
from mo.graph.graph import Graph
|
||||
from mo.middle.replacement import MiddleReplacementPattern
|
||||
from mo.ops.const import Const
|
||||
|
||||
|
||||
class AnchorToPriorBoxes(MiddleReplacementPattern):
|
||||
"""
|
||||
Crop anchors consts before replacing subgraph with all anchors
|
||||
"""
|
||||
enabled = True
|
||||
force_clean_up = True
|
||||
graph_condition = [lambda graph: graph.graph['fw'] == 'mxnet' and graph.graph['cmd_params'].enable_ssd_gluoncv]
|
||||
|
||||
def run_after(self):
|
||||
from extensions.middle.pass_separator import MiddleStart
|
||||
return [MiddleStart]
|
||||
|
||||
def run_before(self):
|
||||
return [SliceLikeToStridedSlice]
|
||||
|
||||
def pattern(self):
|
||||
return dict(
|
||||
nodes=[
|
||||
('const', dict(op='Const')),
|
||||
('const_data', dict(kind='data')),
|
||||
('slice_like', dict(op='slice_like')),
|
||||
('slice_like_out', dict(kind='data')),
|
||||
('reshape', dict(op='Reshape')),
|
||||
],
|
||||
edges=[
|
||||
('const', 'const_data'),
|
||||
('const_data', 'slice_like', {'in': 0}),
|
||||
('slice_like', 'slice_like_out'),
|
||||
('slice_like_out', 'reshape'),
|
||||
]
|
||||
)
|
||||
|
||||
def replace_pattern(self, graph: Graph, match: dict):
|
||||
slice_like = match['slice_like']
|
||||
anchor_node = slice_like.in_port(0).get_source().node
|
||||
reshape = slice_like.out_nodes()[0].out_node()
|
||||
slice_shape = slice_like.out_nodes()[0].shape
|
||||
anchor_node.value = np.copy(anchor_node.value[:slice_shape[0], :slice_shape[1],
|
||||
:slice_shape[2], :slice_shape[3], :slice_shape[4]])
|
||||
anchor_node.shape = slice_shape
|
||||
|
||||
val_node = Const(graph, {'name': slice_like.name +'/croped_', 'value': anchor_node.value[:slice_shape[0], :slice_shape[1],
|
||||
:slice_shape[2], :slice_shape[3], :slice_shape[4]], 'shape': slice_shape}).create_node_with_data()
|
||||
slice_like.in_port(0).disconnect()
|
||||
slice_like.in_port(1).disconnect()
|
||||
slice_like.out_port(0).disconnect()
|
||||
reshape.in_port(0).connect(val_node.in_node().out_port(0))
|
||||
@@ -38,9 +38,9 @@ nodes_attributes = {
|
||||
'split_data_1': {'kind': 'data', 'value': None, 'shape': np.array([1, 227, 227, 1])},
|
||||
'last_data': {'kind': 'data', 'value': None, 'shape': np.array([1, 227, 227, 3])},
|
||||
|
||||
'axis_const': {'kind': 'op'},
|
||||
'axis_const': {'kind': 'op', 'op': 'Const'},
|
||||
'axis_const_data': {'value': np.int64(1), 'shape': None, 'kind': 'data'},
|
||||
'split_dim_const': {'kind': 'op'},
|
||||
'split_dim_const': {'kind': 'op', 'op': 'Const'},
|
||||
'split_dim_const_data': {'value': np.array([1, 2, 3]), 'shape': None, 'kind': 'data'},
|
||||
|
||||
}
|
||||
|
||||
@@ -1,104 +0,0 @@
|
||||
"""
|
||||
Copyright (C) 2018-2020 Intel Corporation
|
||||
|
||||
Licensed under the Apache License, Version 2.0 (the "License");
|
||||
you may not use this file except in compliance with the License.
|
||||
You may obtain a copy of the License at
|
||||
|
||||
http://www.apache.org/licenses/LICENSE-2.0
|
||||
|
||||
Unless required by applicable law or agreed to in writing, software
|
||||
distributed under the License is distributed on an "AS IS" BASIS,
|
||||
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
See the License for the specific language governing permissions and
|
||||
limitations under the License.
|
||||
"""
|
||||
|
||||
import numpy as np
|
||||
|
||||
from extensions.middle.AnchorToPriorBox import AnchorToPriorBoxes
|
||||
from extensions.middle.SliceLikeToStridedSlice import SliceLikeToStridedSlice
|
||||
from mo.graph.graph import Graph
|
||||
from mo.middle.replacement import MiddleReplacementPattern
|
||||
from mo.ops.const import Const
|
||||
|
||||
|
||||
class SsdAnchorsMiddleReplacer(MiddleReplacementPattern):
|
||||
"""
|
||||
Replacing subgraph with all anchors constant to constant op with pre calculated prior boxes values.
|
||||
"""
|
||||
enabled = True
|
||||
force_clean_up = True
|
||||
graph_condition = [lambda graph: graph.graph['fw'] == 'mxnet' and graph.graph['cmd_params'].enable_ssd_gluoncv]
|
||||
|
||||
def run_after(self):
|
||||
return [AnchorToPriorBoxes]
|
||||
|
||||
def run_before(self):
|
||||
return [SliceLikeToStridedSlice]
|
||||
|
||||
|
||||
def pattern(self):
|
||||
return dict(
|
||||
nodes=[
|
||||
('const', dict(op='Const')),
|
||||
('const_data', dict(kind='data')),
|
||||
('reshape0', dict(op='Reshape')),
|
||||
('reshape0_data', dict(kind='data')),
|
||||
('reshape1', dict(op='Reshape')),
|
||||
('reshape1_data', dict(kind='data')),
|
||||
('reshape2', dict(op='Reshape')),
|
||||
('reshape2_data', dict(kind='data')),
|
||||
('reshape3', dict(op='Reshape')),
|
||||
('reshape3_data', dict(kind='data')),
|
||||
('concat', dict(op='Concat')),
|
||||
],
|
||||
edges=[
|
||||
('const', 'const_data'),
|
||||
('const_data', 'reshape0'),
|
||||
('reshape0', 'reshape0_data'),
|
||||
('reshape0_data', 'reshape1'),
|
||||
('reshape1', 'reshape1_data'),
|
||||
('reshape1_data', 'reshape2'),
|
||||
('reshape2', 'reshape2_data'),
|
||||
('reshape2_data', 'reshape3'),
|
||||
('reshape3', 'reshape3_data'),
|
||||
('reshape3_data', 'concat'),
|
||||
])
|
||||
|
||||
def replace_pattern(self, graph: Graph, match: dict):
|
||||
#self.pattern()['nodes']
|
||||
concat_node = match['concat']
|
||||
if len(concat_node.out_nodes()[0].out_nodes()) == 0:
|
||||
return
|
||||
const_values = []
|
||||
for in_node_index in concat_node.in_nodes():
|
||||
current_node = concat_node.in_port(in_node_index).get_source().node
|
||||
for k, v in reversed(self.pattern()['nodes'][:-1]):
|
||||
if 'op' in v:
|
||||
assert current_node.op == v['op']
|
||||
current_node = current_node.in_port(0).get_source().node
|
||||
if current_node.op == 'Const':
|
||||
crop_value = current_node.value
|
||||
crop_value = np.reshape(crop_value, (1, -1))
|
||||
const_values.append(crop_value)
|
||||
break
|
||||
concat_value = np.concatenate(tuple(const_values), axis=1)
|
||||
concat_value = np.reshape(concat_value, (1, 2, -1))
|
||||
slice_value = concat_value[0][0]
|
||||
for i in range(int(concat_value[0][0].size / 4)):
|
||||
index = i * 4
|
||||
xmin = slice_value[index] - (slice_value[index+2] / 2)
|
||||
ymin = slice_value[index + 1] - (slice_value[index + 3] / 2)
|
||||
xmax = slice_value[index] + (slice_value[index + 2] / 2)
|
||||
ymax = slice_value[index + 1] + (slice_value[index + 3] / 2)
|
||||
slice_value[index] = xmin
|
||||
slice_value[index + 1] = ymin
|
||||
slice_value[index + 2] = xmax
|
||||
slice_value[index + 3] = ymax
|
||||
|
||||
val_node = Const(graph, {'name': concat_node.name + '/const_',
|
||||
'value': concat_value}).create_node_with_data()
|
||||
out_node = concat_node.out_port(0).get_destination().node
|
||||
concat_node.out_port(0).disconnect()
|
||||
out_node.in_port(2).connect(val_node.in_node().out_port(0))
|
||||
@@ -50,5 +50,16 @@ class SliceLike(Op):
|
||||
'Input shape ranks are inconsistent: {} and {}'.format(input_shape.size, shape_like.size)
|
||||
node.axes = int64_array(range(shape_like.size))
|
||||
new_shape = np.copy(shape_like)
|
||||
|
||||
node.out_port(0).data.set_shape(new_shape)
|
||||
|
||||
if node.in_port(0).get_connection().data.get_value() is not None:
|
||||
out_value = np.copy(node.in_port(0).data.get_value())
|
||||
|
||||
slice_indexes = []
|
||||
for s in out_value.shape:
|
||||
slice_indexes.append(slice(0, s))
|
||||
|
||||
for axis in node.axes:
|
||||
slice_indexes[axis] = slice(0, new_shape[axis])
|
||||
out_value = out_value[tuple(slice_indexes)]
|
||||
node.out_port(0).data.set_value(out_value)
|
||||
@@ -23,14 +23,18 @@ from mo.graph.graph import Node
|
||||
from mo.utils.unittest.graph import build_graph
|
||||
|
||||
nodes_attributes = {
|
||||
'input_data': {'kind': 'data', 'shape': int64_array([3, 4]), 'value': None},
|
||||
'input': {'kind': 'op', 'op': 'Const'},
|
||||
'input_data': {'kind': 'data', 'shape': int64_array([3, 4]), 'value': np.arange(1, 13).reshape([3, 4])},
|
||||
'shape_like': {'kind': 'op', 'op': 'Const', 'shape': int64_array([2, 3]), 'value': None},
|
||||
'shape_like_data': {'kind': 'data', 'shape': int64_array([2, 3]), 'value': None},
|
||||
'slice_like': {'kind': 'op', 'op': 'slice_data'},
|
||||
'out_data': {'kind': 'data', 'shape': None, 'value': None}
|
||||
}
|
||||
|
||||
edges = [
|
||||
('input', 'input_data'),
|
||||
('input_data', 'slice_like', {'in': 0}),
|
||||
('shape_like', 'shape_like_data'),
|
||||
('shape_like_data', 'slice_like', {'in': 1}),
|
||||
('slice_like', 'out_data')
|
||||
]
|
||||
@@ -43,29 +47,41 @@ class SliceLikeTest(unittest.TestCase):
|
||||
slice_like = Node(graph, 'slice_like')
|
||||
SliceLike.infer(slice_like)
|
||||
ref_shape = int64_array([2, 3])
|
||||
ref_value = np.array([[1, 2, 3], [5, 6, 7]])
|
||||
res_shape = graph.node['out_data']['shape']
|
||||
res_value = graph.node['out_data']['value']
|
||||
self.assertTrue(np.array_equal(res_shape, ref_shape))
|
||||
self.assertTrue(np.array_equal(res_value, ref_value))
|
||||
|
||||
def test_2(self):
|
||||
graph = build_graph(nodes_attributes, edges, {'slice_like': {'axes': (0, 1)}})
|
||||
slice_like = Node(graph, 'slice_like')
|
||||
SliceLike.infer(slice_like)
|
||||
ref_shape = int64_array([2, 3])
|
||||
ref_value = np.array([[1, 2, 3], [5, 6, 7]])
|
||||
res_shape = graph.node['out_data']['shape']
|
||||
res_value = graph.node['out_data']['value']
|
||||
self.assertTrue(np.array_equal(res_shape, ref_shape))
|
||||
self.assertTrue(np.array_equal(res_value, ref_value))
|
||||
|
||||
def test_3(self):
|
||||
graph = build_graph(nodes_attributes, edges, {'slice_like': {'axes': (0,)}})
|
||||
slice_like = Node(graph, 'slice_like')
|
||||
SliceLike.infer(slice_like)
|
||||
ref_shape = int64_array([2, 4])
|
||||
ref_value = np.array([[1, 2, 3, 4], [5, 6, 7, 8]])
|
||||
res_shape = graph.node['out_data']['shape']
|
||||
res_value = graph.node['out_data']['value']
|
||||
self.assertTrue(np.array_equal(res_shape, ref_shape))
|
||||
self.assertTrue(np.array_equal(res_value, ref_value))
|
||||
|
||||
def test_4(self):
|
||||
graph = build_graph(nodes_attributes, edges, {'slice_like': {'axes': (-1,)}})
|
||||
slice_like = Node(graph, 'slice_like')
|
||||
SliceLike.infer(slice_like)
|
||||
ref_shape = int64_array([3, 3])
|
||||
ref_value = np.array([[1, 2, 3], [5, 6, 7], [9, 10, 11]])
|
||||
res_shape = graph.node['out_data']['shape']
|
||||
res_value = graph.node['out_data']['value']
|
||||
self.assertTrue(np.array_equal(res_shape, ref_shape))
|
||||
self.assertTrue(np.array_equal(res_value, ref_value))
|
||||
@@ -448,11 +448,11 @@ class TestOutputCut(unittest.TestCase):
|
||||
# {'embeddings': [{'port': None}]}
|
||||
@generate({'C': [{'port': None}]}, {'C': [{'out': 0}]}, {'C': [{'out': 1}]})
|
||||
def test_output_port_cut(self, output):
|
||||
nodes = {'A': {'type': 'Identity', 'kind': 'op'},
|
||||
'B': {'type': 'Identity', 'kind': 'op'},
|
||||
'C': {'type': 'Identity', 'kind': 'op'},
|
||||
'D': {'type': 'Identity', 'kind': 'op'},
|
||||
'E': {'type': 'Identity', 'kind': 'op'},
|
||||
nodes = {'A': {'type': 'Identity', 'kind': 'op', 'op': 'Identity'},
|
||||
'B': {'type': 'Identity', 'kind': 'op', 'op': 'Identity'},
|
||||
'C': {'type': 'Identity', 'kind': 'op', 'op': 'Identity'},
|
||||
'D': {'type': 'Identity', 'kind': 'op', 'op': 'Identity'},
|
||||
'E': {'type': 'Identity', 'kind': 'op', 'op': 'Identity'},
|
||||
}
|
||||
edges = [
|
||||
('A', 'C', {'in': 0, 'out': 0}),
|
||||
@@ -470,9 +470,9 @@ class TestOutputCut(unittest.TestCase):
|
||||
def test_output_port_cut(self, output):
|
||||
nodes = {'A': {'op': 'Parameter', 'kind': 'op'},
|
||||
'B': {'op': 'Parameter', 'kind': 'op'},
|
||||
'C': {'type': 'Identity', 'kind': 'op'},
|
||||
'D': {'type': 'Identity', 'kind': 'op'},
|
||||
'E': {'type': 'Identity', 'kind': 'op'},
|
||||
'C': {'type': 'Identity', 'kind': 'op', 'op': 'Identity'},
|
||||
'D': {'type': 'Identity', 'kind': 'op', 'op': 'Identity'},
|
||||
'E': {'type': 'Identity', 'kind': 'op', 'op': 'Identity'},
|
||||
}
|
||||
edges = [
|
||||
('A', 'C', {'in': 0, 'out': 0}),
|
||||
@@ -489,9 +489,9 @@ class TestOutputCut(unittest.TestCase):
|
||||
class TestUserDataRepack(unittest.TestCase):
|
||||
nodes = {'A': {'name': 'Aa', 'op': 'Parameter', 'kind': 'op'},
|
||||
'B': {'name': 'Bb', 'op': 'Parameter', 'kind': 'op'},
|
||||
'C': {'name': 'Cc', 'type': 'Identity', 'value': None, 'kind': 'op'},
|
||||
'D': {'name': 'Dd', 'type': 'Identity', 'value': None, 'kind': 'op'},
|
||||
'E': {'name': 'Ee', 'type': 'Identity', 'value': None, 'kind': 'op'},
|
||||
'C': {'name': 'Cc', 'type': 'Identity', 'value': None, 'kind': 'op', 'op': 'Identity'},
|
||||
'D': {'name': 'Dd', 'type': 'Identity', 'value': None, 'kind': 'op', 'op': 'Identity'},
|
||||
'E': {'name': 'Ee', 'type': 'Identity', 'value': None, 'kind': 'op', 'op': 'Identity'},
|
||||
}
|
||||
edges = [
|
||||
('A', 'C', {'in': 0, 'out': 0}),
|
||||
|
||||
@@ -979,7 +979,7 @@ class Graph(nx.MultiDiGraph):
|
||||
undead_node_types.append('TFCustomSubgraphCall')
|
||||
|
||||
if 'cmd_params' in self.graph and getattr(self.graph['cmd_params'], 'keep_shape_ops'):
|
||||
undead_node_types.extend(['ShapeOf', 'Shape'])
|
||||
undead_node_types.extend(['ShapeOf', 'Shape', 'slice_like'])
|
||||
|
||||
mark_output_reachable_nodes(self)
|
||||
shape_inference(self)
|
||||
|
||||
@@ -83,16 +83,14 @@ def mark_undead_nodes(graph, undead_types: list):
|
||||
|
||||
nx.set_node_attributes(G=graph, name='is_undead', values=False)
|
||||
|
||||
# mark output nodes as undead
|
||||
outputs = graph.get_nodes_with_attributes(op='Result')
|
||||
nx.set_node_attributes(G=graph, name='is_undead', values={n: True for n in outputs})
|
||||
undead_types_with_result = undead_types + ['Result']
|
||||
undead_nodes = []
|
||||
for node in graph.get_op_nodes():
|
||||
node_type = node.soft_get('type', node.op)
|
||||
if node_type in undead_types_with_result:
|
||||
undead_nodes.append(node.id)
|
||||
|
||||
# mark specifically defined with node type set of nodes
|
||||
for type in undead_types:
|
||||
node_of_specific_type = graph.get_nodes_with_attributes(type=type)
|
||||
nx.set_node_attributes(G=graph, name='is_undead', values={n: True for n in node_of_specific_type})
|
||||
|
||||
undead_nodes = graph.get_nodes_with_attributes(is_undead=True)
|
||||
nx.set_node_attributes(G=graph, name='is_undead', values={n: True for n in undead_nodes})
|
||||
# propagate 'undead' attribute to children nodes of undead nodes if the node produces constant value
|
||||
for node_name in bfs_search(graph, undead_nodes):
|
||||
if graph.node[node_name]['is_undead']:
|
||||
|
||||
@@ -22,14 +22,14 @@ from mo.graph.graph import Node
|
||||
from mo.middle.passes.eliminate import mark_output_reachable_nodes, mark_const_producer_nodes
|
||||
from mo.utils.unittest.graph import build_graph
|
||||
|
||||
nodes_attributes = {'placeholder_1': {'type': 'Parameter', 'kind': 'op'},
|
||||
'placeholder_2': {'type': 'Parameter', 'kind': 'op'},
|
||||
'node_1': {'type': 'Identity', 'value': None, 'kind': 'op'},
|
||||
'node_2': {'type': 'Identity', 'value': None, 'kind': 'op'},
|
||||
'node_3': {'type': 'Identity', 'value': None, 'kind': 'op'},
|
||||
'node_4': {'type': 'Identity', 'value': None, 'kind': 'op'},
|
||||
'node_5': {'type': 'Identity', 'value': None, 'kind': 'op'},
|
||||
'node_6': {'type': 'Identity', 'value': None, 'kind': 'op'},
|
||||
nodes_attributes = {'placeholder_1': {'type': 'Parameter', 'kind': 'op', 'op': 'Parameter'},
|
||||
'placeholder_2': {'type': 'Parameter', 'kind': 'op', 'op': 'Parameter'},
|
||||
'node_1': {'type': 'Identity', 'value': None, 'kind': 'op', 'op': 'Identity'},
|
||||
'node_2': {'type': 'Identity', 'value': None, 'kind': 'op', 'op': 'Identity'},
|
||||
'node_3': {'type': 'Identity', 'value': None, 'kind': 'op', 'op': 'Identity'},
|
||||
'node_4': {'type': 'Identity', 'value': None, 'kind': 'op', 'op': 'Identity'},
|
||||
'node_5': {'type': 'Identity', 'value': None, 'kind': 'op', 'op': 'Identity'},
|
||||
'node_6': {'type': 'Identity', 'value': None, 'kind': 'op', 'op': 'Identity'},
|
||||
'placeholder_1_data_node': {'value': None, 'kind': 'data'},
|
||||
'placeholder_2_data_node': {'value': None, 'kind': 'data'},
|
||||
'data_node_1': {'value': None, 'kind': 'data'},
|
||||
@@ -39,9 +39,9 @@ nodes_attributes = {'placeholder_1': {'type': 'Parameter', 'kind': 'op'},
|
||||
'data_node_4': {'value': None, 'kind': 'data'},
|
||||
'data_node_5': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'data_node_6': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'tf_call_1': {'type': 'TFCustomSubgraphCall', 'kind': 'op'},
|
||||
'tf_call_2': {'type': 'TFCustomSubgraphCall', 'kind': 'op'},
|
||||
'tf_call_3': {'type': 'TFCustomSubgraphCall', 'kind': 'op'},
|
||||
'tf_call_1': {'type': 'TFCustomSubgraphCall', 'kind': 'op', 'op': 'TFCustomSubgraphCall'},
|
||||
'tf_call_2': {'type': 'TFCustomSubgraphCall', 'kind': 'op', 'op': 'TFCustomSubgraphCall'},
|
||||
'tf_call_3': {'type': 'TFCustomSubgraphCall', 'kind': 'op', 'op': 'TFCustomSubgraphCall'},
|
||||
'op_output': {'kind': 'op', 'op': 'Result'},
|
||||
'op_output_1': {'kind': 'op', 'op': 'Result'},
|
||||
'op_output_2': {'kind': 'op', 'op': 'Result'}
|
||||
|
||||
@@ -29,27 +29,27 @@ nodes_attributes = {
|
||||
'placeholder_2_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
# ScaleShift layer
|
||||
'scaleshift_1': {'type': 'ScaleShift', 'kind': 'op', 'op': 'ScaleShift', 'axis': 0},
|
||||
'const_scaleshift_1_w': {'value': None, 'shape': None, 'kind': 'op'},
|
||||
'const_scaleshift_1_w': {'value': None, 'shape': None, 'kind': 'op', 'op': 'Const'},
|
||||
'scaleshift_1_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_scaleshift_1_b': {'value': None, 'shape': None, 'kind': 'op'},
|
||||
'const_scaleshift_1_b': {'value': None, 'shape': None, 'kind': 'op', 'op': 'Const'},
|
||||
'scaleshift_1_b': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'scaleshift_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
# Mul and Add operations
|
||||
'mul_1': {'type': None, 'value': None, 'kind': 'op', 'op': 'Mul'},
|
||||
'const_mul_1_w': {'value': None, 'shape': None, 'kind': 'op'},
|
||||
'const_mul_1_w': {'value': None, 'shape': None, 'kind': 'op', 'op': 'Const'},
|
||||
'mul_1_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'mul_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'add_1': {'type': None, 'kind': 'op', 'op': 'Add'},
|
||||
'const_add_1_w': {'value': None, 'shape': None, 'kind': 'op'},
|
||||
'const_add_1_w': {'value': None, 'shape': None, 'kind': 'op', 'op': 'Const'},
|
||||
'add_1_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'add_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
# Mul and Add operations
|
||||
'mul_2': {'type': None, 'kind': 'op', 'op': 'Mul'},
|
||||
'const_mul_2_w': {'value': None, 'shape': None, 'kind': 'op'},
|
||||
'const_mul_2_w': {'value': None, 'shape': None, 'kind': 'op', 'op': 'Const'},
|
||||
'mul_2_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'mul_2_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'add_2': {'type': None, 'kind': 'op', 'op': 'Add'},
|
||||
'const_add_2_w': {'value': None, 'shape': None, 'kind': 'op'},
|
||||
'const_add_2_w': {'value': None, 'shape': None, 'kind': 'op', 'op': 'Const'},
|
||||
'add_2_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'add_2_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
# Reshape
|
||||
@@ -59,13 +59,13 @@ nodes_attributes = {
|
||||
'placeholder_2/Reshape_const_data': {'kind': 'data', 'value': None, 'shape': None},
|
||||
# BatchNorm operation
|
||||
'bn_op': {'type': None, 'kind': 'op', 'op': 'BatchNorm', 'can_be_fused': True},
|
||||
'const_bn_const': {'value': None, 'shape': None, 'kind': 'op'},
|
||||
'const_bn_const': {'value': None, 'shape': None, 'kind': 'op', 'op': 'Const'},
|
||||
'bn_const': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_bn_beta': {'value': None, 'shape': None, 'kind': 'op'},
|
||||
'const_bn_beta': {'value': None, 'shape': None, 'kind': 'op', 'op': 'Const'},
|
||||
'bn_beta': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_bn_mean': {'value': None, 'shape': None, 'kind': 'op'},
|
||||
'const_bn_mean': {'value': None, 'shape': None, 'kind': 'op', 'op': 'Const'},
|
||||
'bn_mean': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_bn_var': {'value': None, 'shape': None, 'kind': 'op'},
|
||||
'const_bn_var': {'value': None, 'shape': None, 'kind': 'op', 'op': 'Const'},
|
||||
'bn_var': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'bn_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
# Concat1 operation
|
||||
|
||||
@@ -31,27 +31,27 @@ nodes_attributes = {
|
||||
'scaleshift_1': {'type': 'ScaleShift', 'kind': 'op', 'op': 'ScaleShift'},
|
||||
'scaleshift_1_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'scaleshift_1_b': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_scaleshift_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_scaleshift_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_scaleshift_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'const_scaleshift_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'scaleshift_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
# Mul and Add operations
|
||||
'mul_1': {'type': 'Mul', 'kind': 'op', 'op': 'Mul', 'can_be_fused': True,
|
||||
'infer': lambda node: eltwise_infer(node, lambda a, b: a*b)},
|
||||
'mul_1_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'const_mul_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_mul_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'mul_1_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'add_1': {'type': 'Add', 'kind': 'op', 'op': 'Add', 'can_be_fused': True},
|
||||
'add_1_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'const_add_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_add_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'add_1_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
# Mul2 and Add2 operations
|
||||
'mul_2': {'type': 'Mul', 'kind': 'op', 'op': 'Mul', 'can_be_fused': True},
|
||||
'mul_2_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'const_mul_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_mul_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'mul_2_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'add_2': {'type': 'Add', 'kind': 'op', 'op': 'Add', 'can_be_fused': True},
|
||||
'add_2_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'const_add_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_add_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'add_2_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
# Concat1 operation
|
||||
'concat_1': {'type': 'Concat', 'kind': 'op', 'op': 'Concat'},
|
||||
@@ -60,21 +60,21 @@ nodes_attributes = {
|
||||
'conv_1': {'type': 'Convolution', 'kind': 'op', 'op': 'Conv2D', 'layout': 'NHWC'},
|
||||
'conv_1_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'conv_1_b': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_conv_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_conv_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_conv_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'const_conv_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'conv_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'conv_2': {'type': 'Convolution', 'kind': 'op', 'op': 'Conv2D', 'layout': 'NHWC'},
|
||||
'conv_2_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'conv_2_b': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_conv_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_conv_2_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_conv_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'const_conv_2_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'conv_2_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
# MatMul
|
||||
'fc_1': {'type': 'MatMul', 'kind': 'op', 'layout': 'NHWC'},
|
||||
'fc_1': {'type': 'MatMul', 'kind': 'op', 'layout': 'NHWC', 'op': 'MatMul'},
|
||||
'fc_1_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'fc_1_b': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_fc_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_fc_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_fc_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'const_fc_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'fc_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
# Placeholders
|
||||
'placeholder_2': {'shape': None, 'type': 'Parameter', 'kind': 'op', 'op': 'Parameter'},
|
||||
|
||||
@@ -27,45 +27,45 @@ nodes_attributes = {
|
||||
'placeholder_1_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
# ScaleShift layer
|
||||
'scaleshift_1': {'type': 'ScaleShift', 'kind': 'op', 'op': 'ScaleShift'},
|
||||
'const_scaleshift_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_scaleshift_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'scaleshift_1_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_scaleshift_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_scaleshift_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'scaleshift_1_b': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'scaleshift_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
# Mul and Add operations
|
||||
'mul_1': {'type': 'Mul', 'kind': 'op', 'op': 'Mul', 'can_be_fused': True},
|
||||
'const_mul_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_mul_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'mul_1_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'mul_1_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'add_1': {'type': 'Add', 'kind': 'op', 'op': 'Add', 'can_be_fused': True},
|
||||
'const_add_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_add_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'add_1_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'add_1_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
# Mul2 and Add2 operations
|
||||
'mul_2': {'type': 'Mul', 'kind': 'op', 'op': 'Mul', 'can_be_fused': True},
|
||||
'const_mul_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_mul_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'mul_2_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'mul_2_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'add_2': {'type': 'Add', 'kind': 'op', 'op': 'Add', 'can_be_fused': True},
|
||||
'const_add_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_add_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'add_2_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'add_2_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
# Mul3 and Add3 operations
|
||||
'mul_3': {'type': 'Mul', 'kind': 'op', 'op': 'Mul', 'can_be_fused': True},
|
||||
'const_mul_3_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_mul_3_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'mul_3_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'mul_3_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'add_3': {'type': 'Add', 'kind': 'op', 'op': 'Add', 'can_be_fused': True},
|
||||
'const_add_3_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_add_3_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'add_3_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'add_3_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
# Mul4 and Add4 operations
|
||||
'mul_4': {'type': 'Mul', 'kind': 'op', 'op': 'Mul', 'can_be_fused': True},
|
||||
'const_mul_4_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_mul_4_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'mul_4_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'mul_4_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'add_4': {'type': 'Add', 'kind': 'op', 'op': 'Add', 'can_be_fused': True},
|
||||
'const_add_4_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_add_4_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'add_4_w': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'add_4_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
# Concat1 operation
|
||||
@@ -73,22 +73,22 @@ nodes_attributes = {
|
||||
'concat_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
# Convolutions
|
||||
'conv_1': {'type': 'Convolution', 'kind': 'op', 'op': 'Conv2D', 'layout': 'NHWC'},
|
||||
'const_conv_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_conv_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'conv_1_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_conv_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_conv_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'conv_1_b': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'conv_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'conv_2': {'type': 'Convolution', 'kind': 'op', 'op': 'Conv2D', 'layout': 'NHWC'},
|
||||
'const_conv_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_conv_2_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'conv_2_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_conv_2_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_conv_2_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'conv_2_b': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'conv_2_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
# FullyConnected
|
||||
'fc_1': {'type': 'MatMul', 'kind': 'op', 'op': 'FullyConnected', 'layout': 'NHWC'},
|
||||
'const_fc_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_fc_1_w': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'fc_1_w': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'const_fc_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None},
|
||||
'const_fc_1_b': {'value': None, 'shape': None, 'kind': 'op', 'data_type': None, 'op': 'Const'},
|
||||
'fc_1_b': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
'fc_1_data': {'value': None, 'shape': None, 'kind': 'data'},
|
||||
# Placeholders
|
||||
@@ -96,7 +96,7 @@ nodes_attributes = {
|
||||
'placeholder_2_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'placeholder_3': {'shape': None, 'type': 'Parameter', 'kind': 'op', 'op': 'Parameter'},
|
||||
'placeholder_3_data': {'value': None, 'shape': None, 'kind': 'data', 'data_type': None},
|
||||
'op_output': { 'kind': 'op', 'op': 'Result'}
|
||||
'op_output': {'kind': 'op', 'op': 'Result'}
|
||||
}
|
||||
|
||||
|
||||
|
||||
@@ -277,7 +277,7 @@ regular_op_with_empty_data = lambda name, kwargs: {**regular_op(name, kwargs), *
|
||||
|
||||
# constants
|
||||
const = lambda name, value: {name: {'kind': 'op', 'value': value, 'shape': int64_array(value.shape),
|
||||
'type': 'Const', 'infer': Const.infer}}
|
||||
'type': 'Const', 'infer': Const.infer, 'op': 'Const'}}
|
||||
fake_const = lambda name, shape: {name: {'kind': 'op', 'value': None, 'infer': Const.infer,
|
||||
'shape': int64_array(shape) if shape is not None else None}}
|
||||
shaped_const_with_data = lambda name, shape: {**fake_const(name, shape), **shaped_data(name + '_d', shape)}
|
||||
|
||||
Reference in new issue
Block a user