CenterNet-3D: Optimized for Qualcomm Devices

CenterNet is a machine learning model for generating a birds eye view represenation from the sensors(cameras) mounted on a vehicle.

This is based on the implementation of CenterNet-3D found here. This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.

Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up to run these models on a hosted Qualcomm® device.

Getting Started

There are two ways to deploy this model on your device:

Option 1: Download Pre-Exported Models

Below are pre-exported model assets ready for deployment.

Runtime	Precision	Chipset	SDK Versions	Download
PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® X2 Elite	QAIRT 2.45, ONNX Runtime 1.27.1	Download
PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® X Elite	QAIRT 2.45, ONNX Runtime 1.27.1	Download
PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	QAIRT 2.45, ONNX Runtime 1.27.1	Download
PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	QAIRT 2.45, ONNX Runtime 1.27.1	Download
PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	QAIRT 2.45, ONNX Runtime 1.27.1	Download
PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	QAIRT 2.45, ONNX Runtime 1.27.1	Download
PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite Mobile	QAIRT 2.45, ONNX Runtime 1.27.1	Download
PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	QAIRT 2.45, ONNX Runtime 1.27.1	Download
QNN_CONTEXT_BINARY	w8a16	Snapdragon® X2 Elite	QAIRT 2.45	Download
QNN_CONTEXT_BINARY	w8a16	Snapdragon® X Elite	QAIRT 2.45	Download
QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Gen 3 Mobile	QAIRT 2.45	Download
QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Gen 1 Mobile	QAIRT 2.45	Download
QNN_CONTEXT_BINARY	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	QAIRT 2.45	Download
QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA8775P	QAIRT 2.45	Download
QNN_CONTEXT_BINARY	w8a16	Qualcomm® Dragonwing™ IQ-9075	QAIRT 2.45	Download
QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA7255P	QAIRT 2.45	Download
QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA8295P	QAIRT 2.45	Download
QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite Mobile	QAIRT 2.45	Download
QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit CenterNet-3D on Qualcomm® AI Hub.

Option 2: Export with Custom Configurations

Use the Qualcomm® AI Hub Models Python library to compile and export the model with your own:

• Custom weights (e.g., fine-tuned checkpoints)
• Custom input shapes
• Target device and runtime configurations

This option is ideal if you need to customize the model beyond the default configuration provided here.

See our repository for CenterNet-3D on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.driver_assistance

Model Stats:

• Model checkpoint: ddd_3dop.pth
• Input resolution: 1 x 3 x 384 x 1280
• Number of parameters: 20.6M
• Model size: 79 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
CenterNet-3D	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® X2 Elite	2001.902 ms	5 - 5 MB	NPU
CenterNet-3D	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® X Elite	2681.5 ms	57 - 57 MB	NPU
CenterNet-3D	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	2376.278 ms	4 - 16 MB	NPU
CenterNet-3D	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	4469.878 ms	3 - 17 MB	NPU
CenterNet-3D	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	3016.311 ms	1 - 63 MB	NPU
CenterNet-3D	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® QCS8450	4469.878 ms	3 - 17 MB	NPU
CenterNet-3D	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	2863.674 ms	0 - 6 MB	NPU
CenterNet-3D	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	2681.5 ms	57 - 57 MB	NPU
CenterNet-3D	PRECOMPILED_QNN_ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	2388.767 ms	3 - 14 MB	NPU
CenterNet-3D	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite Mobile	2388.767 ms	3 - 14 MB	NPU
CenterNet-3D	PRECOMPILED_QNN_ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	2742.523 ms	1 - 11 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Snapdragon® X2 Elite	1999.504 ms	3 - 3 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Snapdragon® X Elite	2724.356 ms	3 - 3 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Gen 3 Mobile	2367.036 ms	3 - 12 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Gen 1 Mobile	4469.269 ms	3 - 12 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Qualcomm® Dragonwing™ QCS8275	3849.673 ms	1 - 10 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	3004.329 ms	4 - 5 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA8775P	2901.894 ms	2 - 10 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA8650P	2901.894 ms	2 - 10 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA8255P	2901.894 ms	2 - 10 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Qualcomm® QCS8450	4469.269 ms	3 - 12 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Qualcomm® Dragonwing™ IQ-9075	2863.035 ms	3 - 8 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Qualcomm® Dragonwing™ IQ-X7181	2724.356 ms	3 - 3 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Qualcomm® Dragonwing™ Q-8750	2346.026 ms	1 - 10 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA7255P	3849.673 ms	1 - 10 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Qualcomm® SA8295P	3466.707 ms	4 - 9 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite Mobile	2346.026 ms	1 - 10 MB	NPU
CenterNet-3D	QNN_CONTEXT_BINARY	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	2740.049 ms	0 - 9 MB	NPU

License

• The license for the original implementation of CenterNet-3D can be found here.

References

• Objects as Points
• Source Model Implementation

Community

• Join our AI Hub Slack community to collaborate, post questions and learn more about on-device AI.
• For questions or feedback please reach out to us.