DeformableDETR: Optimized for Qualcomm Devices

Deformable DETR is a machine learning model that can detect objects (trained on COCO dataset).

This is based on the implementation of DeformableDETR 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
ONNX	float	Universal	QAIRT 2.42, ONNX Runtime 1.24.1	Download
ONNX	w8a16	Universal	QAIRT 2.42, ONNX Runtime 1.24.1	Download

For more device-specific assets and performance metrics, visit DeformableDETR 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 DeformableDETR on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.object_detection

Model Stats:

• Model checkpoint: deformable-detr
• Input resolution: 480x480
• Number of parameters: 40M
• Model size: 160 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
DeformableDETR	ONNX	w8a16	Snapdragon® X2 Elite	1632.94 ms	93 - 93 MB	NPU
DeformableDETR	ONNX	w8a16	Snapdragon® X Elite	2795.423 ms	90 - 90 MB	NPU
DeformableDETR	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	2355.274 ms	67 - 2032 MB	NPU
DeformableDETR	ONNX	w8a16	Qualcomm® QCS6490	7549.381 ms	1052 - 1059 MB	CPU
DeformableDETR	ONNX	w8a16	Qualcomm® QCS9075	3049.633 ms	62 - 67 MB	NPU
DeformableDETR	ONNX	w8a16	Qualcomm® QCM6690	4179.147 ms	1030 - 1052 MB	CPU
DeformableDETR	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	1684.91 ms	63 - 1330 MB	NPU
DeformableDETR	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	3776.96 ms	1044 - 1067 MB	CPU
DeformableDETR	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	1412.598 ms	65 - 1358 MB	NPU

License

• The license for the original implementation of DeformableDETR can be found here.

References

• Deformable DETR: Deformable Transformers for End-to-End Object Detection
• 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.