Model Demo: Centernet Detection

NOTE: The Jupyter Notebook below is included in the Chimera SDK and can be run interactively by running the following CLI command:

$ quadric sdk notebook

From the Jupyter Notebook window in your browser, select the notebook named /quadric/sdk-cli/examples/models/centernet/centernet.ipynb.

from sdk_cli.visualizers.centernet import CenternetVisualizer
from sdk_cli.node_builtins.outputs.centernet_visualization import centernet

c = CenternetVisualizer(centernet.LABELS["VOCO"])

Length of classes: 20

import onnx
from sdk_cli.utils import model_helpers
import onnxruntime as ort
import numpy as np
import matplotlib.pyplot as plt

fp_onnx_name = f"centernet_float32.onnx"
from tvm.contrib.epu.chimera_job.quantize import quadric_quantize

images_path = "../../common/calibration/coco-like"
## include quadric's cli helpers and instantiate a module to help

model = onnx.load(fp_onnx_name)
_input = model.graph.input[0]
_input_shape = [dim_value.dim_value for dim_value in _input.type.tensor_type.shape.dim]
print(f"NCHW: {_input_shape}")
dataset_input_size = (_input_shape[-1], _input_shape[-2])  # an (W, H) tuple

dataset_mean = [0.5, 0.5, 0.5]
dataset_std = [0.5, 0.5, 0.5]
mh = model_helpers.ModelHelper(dataset_input_size, dataset_mean, dataset_std)

NCHW: [1, 3, 544, 544]


/tmp/ipykernel_2662/1708068201.py:21: DeprecationWarning: Call to deprecated class ModelHelper. ('ModelHelper' class is being deprecated. Quadric APIs have been updated to use PyTorch datasets and transforms instead.) -- Deprecated since version 24.01.
  mh = model_helpers.ModelHelper(dataset_input_size, dataset_mean, dataset_std)

quantize_result = quadric_quantize(fp_onnx_name, 100, mh, images_path)

2026-07-18 12:09 - INFO - epu - quantize - Collecting calibration data
2026-07-18 12:09 - INFO - epu - quantize - Optimized model to opset
2026-07-18 12:09 - INFO - epu - quantize - Converted model to opset 12
2026-07-18 12:09 - INFO - epu - quantize - Saved optimized model to centernet_float32_float32_opt.onnx
2026-07-18 12:09 - INFO - epu - quantize - Input shapes: [1, 3, 544, 544]. Input names: input
2026-07-18 12:09 - INFO - epu - quantize - Output shapes: [[1, 20, 136, 136], [1, 2, 136, 136], [1, 2, 136, 136]]. Output names: ['output', '387', '390']
2026-07-18 12:09 - INFO - epu - quantize - applying calibration data to input: input
2026-07-18 12:09 - INFO - epu - quantize - calibration set size: 9
2026-07-18 12:09 - INFO - epu - quantize - Running real quantization on this input: input with input shape: [1, 3, 544, 544]
2026-07-18 12:09 - DEBUG - epu - quantize - Full exclusion set for quantization: ['Softmax', 'Sigmoid', 'QuadricCustomOp']
2026-07-18 12:09 - DEBUG - epu - quantize - excl_nodes ['Sigmoid_98']
2026-07-18 12:09 - INFO - epu - quantize - Quantization started...
WARNING:root:Please use QuantFormat.QDQ for activation type QInt8 and weight type QInt8. Or it will lead to bad performance on x64.
2026-07-18 12:09 - INFO - epu - quantize - Quantization done succesfully!
2026-07-18 12:09 - INFO - epu - quantize - ONNX full precision model size: 102.27 MB
2026-07-18 12:09 - INFO - epu - quantize - ONNX quantized model size: 25.65 MB
2026-07-18 12:09 - INFO - epu - quantize - Saved quantized model to /quadric/sdk-cli/examples/models/centernet/centernet_opt_sym_int8_q.onnx
2026-07-18 12:09 - INFO - epu - quantize - Saved shape inferenced model to /quadric/sdk-cli/examples/models/centernet/centernet_opt_sym_int8_q.onnx
2026-07-18 12:09 - INFO - epu - quantize - Checking for remaining FLOAT/FLOAT16 types.
2026-07-18 12:09 - INFO - epu - quantize - Model still has FLOAT/FLOAT16 types. Creating ranges for floating point tensors using calibration data


Custom quantization code for ConvTranspose
Custom quantization code for ConvTranspose
Custom quantization code for ConvTranspose


2026-07-18 12:10 - INFO - epu - quantize - Saved tensor ranges to /quadric/sdk-cli/examples/models/centernet/centernet_opt_sym_int8_q.onnx.tranges

def run_onnx_all_images(images_path, onnx_name):
    vis_image_list = []
    allimages = mh.get_images(images_path)
    ort_sess = ort.InferenceSession(onnx_name)
    for image_p in allimages:
        image = mh.load_image(image_p)
        onnx_output = ort_sess.run([], {"input": image})
        output_img = CenternetVisualizer.input_img_to_display(image)
        vis_img = c.draw_boxes(onnx_output, output_img, threshold=0.5, scale=4)
        vis_image_list.append(vis_img)
    return vis_image_list


list_fp = run_onnx_all_images(images_path, fp_onnx_name)
print(list_fp)

[<PIL.PngImagePlugin.PngImageFile image mode=RGBA size=640x480 at 0x70F7CEDB8BB0>, <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=640x480 at 0x70F7CF58A290>, <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=640x480 at 0x70F7CF5E3EE0>, <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=640x480 at 0x70F7CF63FF10>, <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=640x480 at 0x70F7CF4CF130>, <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=640x480 at 0x70F7CF51BEB0>, <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=640x480 at 0x70F7CF39EC20>, <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=640x480 at 0x70F7CF3EEA70>, <PIL.PngImagePlugin.PngImageFile image mode=RGBA size=640x480 at 0x70F7CF453D00>]

from tvm.contrib.epu.chimera_job.chimera_job import ChimeraJob

cgc_job = ChimeraJob(model_p=quantize_result.qmodel_path, trange_file=quantize_result.tranges_path)
cgc_job.compile(quiet=True)

THREADS = 2

all_images = mh.get_images(images_path, THREADS)
all_inputs = [{"input": mh.load_image(image)} for image in all_images]
all_results = cgc_job.run_batch_inference_harness(inputs=all_inputs, threads=THREADS)

Processing: 100%|███████████████████████████████| 2/2 [05:06<00:00, 153.03s/it]

for i, result in enumerate(all_results):
    image = mh.load_image(all_images[i])
    output_img = CenternetVisualizer.input_img_to_display(image)
    list_output = [result["output"], result["387"], result["390"]]
    img = c.draw_boxes(list_output, output_img, threshold=0.5, scale=4)
plt.show()

print(cgc_job)
cgc_job.plot_run_statistics();

╒═════════════════════╤══════════════════════════════════════════════════════════════════════════╕
│ Module Name         │ centernet_opt_sym_int8_q_QC_U_1d7_16MB_4kB_128GBps_128GBps_16_OFF_x1_x1  │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ ONNX File           │ /quadric/sdk-cli/examples/models/centernet/centernet_opt_sym_int8_q.onnx │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ Product Target      │ QC-U                                                                     │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ Number of Cores     │ 1                                                                        │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ ISS Clock Frequency │ 1.700                                                                    │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ L2M Size            │ 16MB                                                                     │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ LRM Size            │ 4kB                                                                      │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ External Read BW    │ 128GBps                                                                  │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ External Write BW   │ 128GBps                                                                  │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ MACS per PE         │ 16                                                                       │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ Max L2M             │ 15.267MB                                                                 │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ Max LRM             │ 2.750kB                                                                  │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ Max Temp Ext Bytes  │ 0.000MB                                                                  │
├─────────────────────┼──────────────────────────────────────────────────────────────────────────┤
│ Network GMACs       │ 52.861                                                                   │
╘═════════════════════╧══════════════════════════════════════════════════════════════════════════╛

╒════╤════════╤════════╤═══════════════════╤══════════════════════════╤═══════╕
│    │ Type   │ Name   │ shape             │ type                     │ mse   │
╞════╪════════╪════════╪═══════════════════╪══════════════════════════╪═══════╡
│  0 │ Input  │ input  │ [1, 3, 544, 544]  │ tensor[FixedPoint32<30>] │ n/a   │
├────┼────────┼────────┼───────────────────┼──────────────────────────┼───────┤
│  1 │ Output │ output │ [1, 20, 136, 136] │ tensor[FixedPoint32<31>] │ 0.000 │
├────┼────────┼────────┼───────────────────┼──────────────────────────┼───────┤
│  2 │ Output │ 387    │ [1, 2, 136, 136]  │ tensor[FixedPoint32<24>] │ 1.197 │
├────┼────────┼────────┼───────────────────┼──────────────────────────┼───────┤
│  3 │ Output │ 390    │ [1, 2, 136, 136]  │ tensor[FixedPoint32<30>] │ 0.000 │
╘════╧════════╧════════╧═══════════════════╧══════════════════════════╧═══════╛

Post-ISS Report 1.7 GHz ***
Fully placed-and-routed gate simulation: 
╒══════════════════════════════════╤═════════╕
│ Latency (ms)                     │ 5.76    │
├──────────────────────────────────┼─────────┤
│ FPS                              │ 173.55  │
├──────────────────────────────────┼─────────┤
│ Average Power @ 3nm SSGNP (mW)   │ 1821.44 │
├──────────────────────────────────┼─────────┤
│ FPS per Watt @ 3nm SSGNP (FPS/W) │ 95.28   │
├──────────────────────────────────┼─────────┤
│ Ext Rd Bytes (MB)                │ 29.10   │
├──────────────────────────────────┼─────────┤
│ Ext Wr Bytes (MB)                │ 1.69    │
├──────────────────────────────────┼─────────┤
│ Avg Ext Rd BW (GBps)             │ 4.93    │
├──────────────────────────────────┼─────────┤
│ Avg Ext Wr BW (GBps)             │ 0.29    │
├──────────────────────────────────┼─────────┤
│ MAC Utilization                  │ 32.94%  │
╘══════════════════════════════════╧═════════╛
*** Data generated using 7nm SSGNP gatesim and scaled to 3nm

[SDK-CLI] : TotalCycles: 9,795,489
[SDK-CLI] : Executions/second: 173.55

compute      : ▇▇▇▇▇▇▇▇ 1.179M
data_array   : ▇▇▇▇▇▇ 855.04K
mac          : ▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 7.051M
data_external: ▏ 68.375K
data_ocm     : ▇▇▇ 526.174K

for more information check run directory: /quadric/sdk-cli/examples/models/centernet/ccl_build/centernet_opt_sym_int8_q_QC_U_1d7_16MB_4kB_128GBps_128GBps_16_OFF_x1_x1/run/20260718_121300_9a5307


2026-07-18 12:18 - INFO - epu - chimera_job - Combined plots generated and saved to: 
/quadric/sdk-cli/examples/models/centernet/ccl_build/centernet_opt_sym_int8_q_QC_U_1d7_16MB_4kB_128GBps_128GBps_16_OFF_x1_x1/run/20260718_121300_9a5307/data/centernet_opt_sym_int8_q_QC_U_1d7_16MB_4kB_128GBps_128GBps_16_OFF_x1_x1.combined.png