diff --git a/src/finn/custom_op/im2col.py b/src/finn/custom_op/im2col.py
index 0e6281d193d3d5981b48316aff567567b477bb55..e2fe918ab1b84a486a238faad33e009fac192973 100644
--- a/src/finn/custom_op/im2col.py
+++ b/src/finn/custom_op/im2col.py
@@ -103,7 +103,7 @@ class Im2Col(CustomOp):
return helper.make_node(
"Constant",
inputs=[],
- outputs=["values"],
+ outputs=[self.onnx_node.output[0]],
value=helper.make_tensor(
name="const_tensor",
data_type=TensorProto.FLOAT,
diff --git a/src/finn/transformation/lower_convs_to_matmul.py b/src/finn/transformation/lower_convs_to_matmul.py
new file mode 100644
index 0000000000000000000000000000000000000000..ed8fd21d812dea494e135a7967d8bb7f9aae30b5
--- /dev/null
+++ b/src/finn/transformation/lower_convs_to_matmul.py
@@ -0,0 +1,126 @@
+# Copyright (c) 2020, Xilinx
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of FINN nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+from onnx import TensorProto
+from onnx import helper
+
+from finn.transformation import Transformation
+from finn.transformation.infer_shapes import InferShapes
+from finn.util.basic import get_by_name
+
+
+class LowerConvsToMatMul(Transformation):
+ """Replace Conv layers with pairs of Im2Col-MatMul layers, plus Transpose
+ layers to keep the original data layout."""
+
+ def apply(self, model):
+ graph = model.graph
+ node_ind = 0
+ graph_modified = False
+ for n in graph.node:
+ node_ind += 1
+ if n.op_type == "Conv":
+ graph_modified = True
+ cnv_input = n.input[0]
+ cnv_output = n.output[0]
+ idt = model.get_tensor_datatype(cnv_input)
+ odt = model.get_tensor_datatype(cnv_output)
+ # extract conv parameters
+ k = get_by_name(n.attribute, "kernel_shape").ints[-1]
+ pad = get_by_name(n.attribute, "pads").ints[-1]
+ stride = get_by_name(n.attribute, "strides").ints[-1]
+ weight_name = n.input[1]
+ W_conv = model.get_initializer(weight_name)
+ ifm_ch = W_conv.shape[1]
+ ofm_ch = W_conv.shape[0]
+ ifm_dim = model.get_tensor_shape(n.input[0])[-1] # assume NCHW
+ ofm_dim = model.get_tensor_shape(n.output[0])[-1] # assume NCHW
+ # reuse conv weights for new matmul weights
+ W_matmul = W_conv.reshape(ofm_ch, ifm_ch * k * k).T
+ model.set_initializer(weight_name, W_matmul)
+ # create new intermediate values
+ inp_trans_out = helper.make_tensor_value_info(
+ model.make_new_valueinfo_name(),
+ TensorProto.FLOAT,
+ (1, ifm_dim, ifm_dim, ifm_ch), # NHWC
+ )
+ graph.value_info.append(inp_trans_out)
+ inp_trans_out = inp_trans_out.name
+ model.set_tensor_datatype(inp_trans_out, idt)
+
+ im2col_out = helper.make_tensor_value_info(
+ model.make_new_valueinfo_name(),
+ TensorProto.FLOAT,
+ (1, ofm_dim, ofm_dim, ifm_ch * k * k),
+ )
+ graph.value_info.append(im2col_out)
+ im2col_out = im2col_out.name
+ model.set_tensor_datatype(im2col_out, idt)
+
+ matmul_out = helper.make_tensor_value_info(
+ model.make_new_valueinfo_name(),
+ TensorProto.FLOAT,
+ (1, ofm_dim, ofm_dim, ofm_ch),
+ )
+ graph.value_info.append(matmul_out)
+ matmul_out = matmul_out.name
+ model.set_tensor_datatype(matmul_out, odt)
+
+ # create new nodes
+ # NCHW -> NHWC
+ inp_trans_node = helper.make_node(
+ "Transpose", [cnv_input], [inp_trans_out], perm=[0, 2, 3, 1]
+ )
+ # lower input tensor
+ im2col_node = helper.make_node(
+ "Im2Col",
+ [inp_trans_out],
+ [im2col_out],
+ domain="finn",
+ stride=stride,
+ kernel_size=k,
+ pad_amount=pad,
+ input_shape="(1,{},{},{})".format(ifm_dim, ifm_dim, ifm_ch),
+ )
+ # do matmul
+ matmul_node = helper.make_node(
+ "MatMul", [im2col_out, weight_name], [matmul_out]
+ )
+ # NHWC -> NCHW
+ out_trans_node = helper.make_node(
+ "Transpose", [matmul_out], [cnv_output], perm=[0, 3, 1, 2]
+ )
+ # insert nodes where the conv is to preserve topological ordering
+ graph.node.insert(node_ind, inp_trans_node)
+ graph.node.insert(node_ind + 1, im2col_node)
+ graph.node.insert(node_ind + 2, matmul_node)
+ graph.node.insert(node_ind + 3, out_trans_node)
+ # remove old nodes
+ graph.node.remove(n)
+ model = model.transform(InferShapes())
+ return (model, graph_modified)
diff --git a/tests/transformation/test_conv_lowering.py b/tests/transformation/test_conv_lowering.py
new file mode 100644
index 0000000000000000000000000000000000000000..85dd0f721593b9f91796e434a53e242800d73edd
--- /dev/null
+++ b/tests/transformation/test_conv_lowering.py
@@ -0,0 +1,63 @@
+# Copyright (c) 2020, Xilinx
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are met:
+#
+# * Redistributions of source code must retain the above copyright notice, this
+# list of conditions and the following disclaimer.
+#
+# * Redistributions in binary form must reproduce the above copyright notice,
+# this list of conditions and the following disclaimer in the documentation
+# and/or other materials provided with the distribution.
+#
+# * Neither the name of FINN nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+import os
+import pkg_resources as pk
+import brevitas.onnx as bo
+import numpy as np
+
+
+from finn.core.modelwrapper import ModelWrapper
+from finn.transformation.fold_constants import FoldConstants
+from finn.transformation.infer_shapes import InferShapes
+from finn.util.test import get_test_model_trained
+from finn.transformation.lower_convs_to_matmul import LowerConvsToMatMul
+import finn.core.onnx_exec as oxe
+
+export_onnx_path = "test_output_cnv.onnx"
+
+
+def test_conv_lowering_cnv_w1a1():
+ cnv = get_test_model_trained("CNV", 1, 1)
+ bo.export_finn_onnx(cnv, (1, 3, 32, 32), export_onnx_path)
+ model = ModelWrapper(export_onnx_path)
+ model = model.transform(InferShapes())
+ model = model.transform(FoldConstants())
+ fn = pk.resource_filename("finn", "data/cifar10/cifar10-test-data-class3.npz")
+ input_tensor = np.load(fn)["arr_0"].astype(np.float32)
+ assert input_tensor.shape == (1, 3, 32, 32)
+ # execute imported model to get expected answer
+ input_dict = {"0": input_tensor}
+ output_dict_e = oxe.execute_onnx(model, input_dict)
+ expected = output_dict_e[list(output_dict_e.keys())[0]]
+ # execute transformed model and compare
+ model = model.transform(LowerConvsToMatMul())
+ output_dict_p = oxe.execute_onnx(model, input_dict)
+ produced = output_dict_p[list(output_dict_p.keys())[0]]
+ assert np.isclose(produced, expected).all()
+ os.remove(export_onnx_path)