Add German Traffic Sign Recognition CNN to finn-examples

build/gtsrb/build.py

@@ -0,0 +1,125 @@
+# Copyright (C) 2024, Advanced Micro Devices, Inc.
+# 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 finn.builder.build_dataflow as build
+import finn.builder.build_dataflow_config as build_cfg
+from finn.builder.build_dataflow_config import default_build_dataflow_steps
+from qonnx.core.datatype import DataType
+import os
+import shutil
+import numpy as np
+from onnx import helper as oh
+
+models = [
+    "cnv_1w1a_gtsrb",
+]
+
+# which platforms to build the networks for
+zynq_platforms = ["Pynq-Z1"]
+platforms_to_build = zynq_platforms
+
+
+def custom_step_add_preproc(model, cfg):
+    # GTSRB data with raw uint8 pixels is divided by 255 prior to training
+    # reflect this in the inference graph so we can perform inference directly
+    # on raw uint8 data
+    in_name = model.graph.input[0].name
+    new_in_name = model.make_new_valueinfo_name()
+    new_param_name = model.make_new_valueinfo_name()
+    div_param = np.asarray(255.0, dtype=np.float32)
+    new_div = oh.make_node(
+        "Div",
+        [in_name, new_param_name],
+        [new_in_name],
+        name="PreprocDiv",
+    )
+    model.set_initializer(new_param_name, div_param)
+    model.graph.node.insert(0, new_div)
+    model.graph.node[1].input[0] = new_in_name
+    # set input dtype to uint8
+    model.set_tensor_datatype(in_name, DataType["UINT8"])
+    return model
+
+
+custom_build_steps = [custom_step_add_preproc] + default_build_dataflow_steps
+
+
+# determine which shell flow to use for a given platform
+def platform_to_shell(platform):
+    if platform in zynq_platforms:
+        return build_cfg.ShellFlowType.VIVADO_ZYNQ
+    else:
+        raise Exception("Unknown platform, can't determine ShellFlowType")
+
+
+# create a release dir, used for finn-examples release packaging
+os.makedirs("release", exist_ok=True)
+
+for platform_name in platforms_to_build:
+    shell_flow_type = platform_to_shell(platform_name)
+    vitis_platform = None
+    # for Zynq, use the board name as the release name
+    # e.g. ZCU104
+    release_platform_name = platform_name
+    platform_dir = "release/%s" % release_platform_name
+    os.makedirs(platform_dir, exist_ok=True)
+    for model_name in models:
+        # set up the build configuration for this model
+        cfg = build_cfg.DataflowBuildConfig(
+            output_dir="output_%s_%s" % (model_name, release_platform_name),
+            target_fps=3000,
+            synth_clk_period_ns=10.0,
+            board=platform_name,
+            steps=custom_build_steps,
+            folding_config_file="folding_config/cnv_gtsrb_folding_config.json",
+            shell_flow_type=shell_flow_type,
+            vitis_platform=vitis_platform,
+            generate_outputs=[
+                build_cfg.DataflowOutputType.ESTIMATE_REPORTS,
+                build_cfg.DataflowOutputType.STITCHED_IP,
+                build_cfg.DataflowOutputType.RTLSIM_PERFORMANCE,
+                build_cfg.DataflowOutputType.BITFILE,
+            ],
+            save_intermediate_models=True,
+            default_swg_exception=True,
+        )
+        model_file = "models/%s.onnx" % model_name
+        # launch FINN compiler to build
+        build.build_dataflow_cfg(model_file, cfg)
+        # copy bitfiles into release dir if found
+        bitfile_gen_dir = cfg.output_dir + "/bitfile"
+        files_to_check_and_copy = [
+            "finn-accel.bit",
+            "finn-accel.hwh",
+            "finn-accel.xclbin",
+        ]
+        for f in files_to_check_and_copy:
+            src_file = bitfile_gen_dir + "/" + f
+            dst_file = platform_dir + "/" + f.replace("finn-accel", model_name)
+            if os.path.isfile(src_file):
+                shutil.copy(src_file, dst_file)

build/gtsrb/folding_config/cnv_gtsrb_folding_config.json

@@ -0,0 +1,78 @@
+{
+    "Defaults": {},
+    "Thresholding_rtl_0": {
+      "PE": 1
+    },
+    "ConvolutionInputGenerator_rtl_0": {
+      "SIMD": 3,
+      "ram_style": "distributed"
+    },
+    "MVAU_hls_0": {
+      "PE": 16,
+      "SIMD": 3,
+      "ram_style": "auto"
+    },
+    "ConvolutionInputGenerator_rtl_1": {
+      "SIMD": 32,
+      "ram_style": "distributed"
+    },
+    "MVAU_hls_1": {
+      "PE": 32,
+      "SIMD": 32,
+      "ram_style": "auto"
+    },
+    "ConvolutionInputGenerator_rtl_2": {
+      "SIMD": 32,
+      "ram_style": "distributed"
+    },
+    "MVAU_hls_2": {
+      "PE": 16,
+      "SIMD": 32,
+      "ram_style": "auto"
+    },
+    "ConvolutionInputGenerator_rtl_3": {
+      "SIMD": 32,
+      "ram_style": "distributed"
+    },
+    "MVAU_hls_3": {
+      "PE": 16,
+      "SIMD": 32,
+      "ram_style": "auto"
+    },
+    "ConvolutionInputGenerator_rtl_4": {
+      "SIMD": 32,
+      "ram_style": "distributed"
+    },
+    "MVAU_hls_4": {
+      "PE": 4,
+      "SIMD": 32,
+      "ram_style": "auto"
+    },
+    "ConvolutionInputGenerator_rtl_5": {
+      "SIMD": 32,
+      "ram_style": "distributed"
+    },
+    "MVAU_hls_5": {
+      "PE": 1,
+      "SIMD": 32,
+      "ram_style": "auto"
+    },
+    "MVAU_hls_6": {
+      "PE": 1,
+      "SIMD": 4,
+      "ram_style": "auto"
+    },
+    "MVAU_hls_7": {
+      "PE": 1,
+      "SIMD": 8,
+      "ram_style": "auto"
+    },
+    "MVAU_hls_8": {
+      "PE": 4,
+      "SIMD": 1,
+      "ram_style": "auto"
+    },
+    "LabelSelect_hls_0": {
+      "PE": 1
+    }
+}

build/gtsrb/models/download-model.sh

@@ -0,0 +1,2 @@
+#!/bin/bash
+wget https://github.com/fastmachinelearning/qonnx_model_zoo/raw/feature/gtsrb_cnv/models/GTSRB/Brevitas_CNV1W1A/cnv_1w1a_gtsrb.onnx

ci/Jenkinsfile

@@ -23,7 +23,8 @@ pipeline {
                              "kws",
                              "mobilenet-v1",
                              "resnet50",
-                             "vgg10-radioml"]
+                             "vgg10-radioml",
+                             "gtsrb"]
             createParallelBuilds(buildList)
             createReleaseArea(buildList)
           }

finn_examples/models.py

@@ -67,6 +67,17 @@
     "num_outputs": 1,
 }
 
+_gtsrb_cnv_io_shape_dict = {
+    "idt": DataType["UINT8"],
+    "odt": DataType["INT16"],
+    "ishape_normal": (1, 32, 32, 3),
+    "oshape_normal": (1, 44),
+    "ishape_folded": (1, 1, 32, 32, 3, 1),
+    "oshape_folded": (1, 11, 4),
+    "ishape_packed": (1, 1, 32, 32, 3, 1),
+    "oshape_packed": (1, 11, 8),
+}
+
 _bincop_cnv_io_shape_dict = {
     "idt": [DataType["UINT8"]],
     "odt": [DataType["UINT8"]],
@@ -352,7 +363,6 @@ def resnet50_w1a2_imagenet(target_platform=None):
         runtime_weight_dir=runtime_weight_dir,
     )
 
-
 def vgg10_w4a4_radioml(target_platform=None):
     target_platform = resolve_target_platform(target_platform)
     driver_mode = get_driver_mode()
@@ -366,7 +376,6 @@ def vgg10_w4a4_radioml(target_platform=None):
         fclk_mhz=fclk_mhz,
     )
 
-
 def mlp_w2a2_unsw_nb15(target_platform=None):
     target_platform = resolve_target_platform(target_platform)
     driver_mode = get_driver_mode()
@@ -376,3 +385,10 @@ def mlp_w2a2_unsw_nb15(target_platform=None):
     return FINNExampleOverlay(
         filename, driver_mode, _unsw_nb15_mlp_io_shape_dict, fclk_mhz=fclk_mhz
     )
+
+def cnv_w1a1_gtsrb(target_platform=None):
+    target_platform = resolve_target_platform(target_platform)
+    driver_mode = get_driver_mode()
+    model_name = "cnv-gtsrb-w1a1"
+    filename = find_bitfile(model_name, target_platform)
+    return FINNExampleOverlay(filename, driver_mode, _gtsrb_cnv_io_shape_dict)