DETR-ResNet50-DC5: Optimized for Qualcomm Devices

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

This is based on the implementation of DETR-ResNet50-DC5 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 2.45, ONNX Runtime 1.27.1	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
TFLITE	float	Universal	QAIRT 2.45	Download

For more device-specific assets and performance metrics, visit DETR-ResNet50-DC5 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 DETR-ResNet50-DC5 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.object_detection

Model Stats:

• Model checkpoint: ResNet50-DC5
• Input resolution: 480x480
• Model size (float): 160 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
DETR-ResNet50-DC5	ONNX	float	Snapdragon® X2 Elite	19.928 ms	5 - 5 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® X Elite	44.735 ms	78 - 78 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® 8 Gen 3 Mobile	31.663 ms	3 - 488 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® 8 Gen 1 Mobile	68.257 ms	1 - 413 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	42.989 ms	0 - 93 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS8450	68.257 ms	1 - 413 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	57.372 ms	5 - 12 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	44.735 ms	78 - 78 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® Dragonwing™ Q-8750	23.038 ms	3 - 374 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® 8 Elite Mobile	23.038 ms	3 - 374 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	18.419 ms	3 - 386 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® X2 Elite	21.283 ms	5 - 5 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® X Elite	46.549 ms	5 - 5 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	32.528 ms	3 - 467 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	71.752 ms	4 - 396 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8275	173.118 ms	1 - 371 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	44.672 ms	5 - 7 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA8775P	57.788 ms	1 - 371 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA8650P	57.788 ms	1 - 371 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA8255P	57.788 ms	1 - 371 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8450	71.752 ms	4 - 396 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	58.587 ms	5 - 11 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	46.549 ms	5 - 5 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	23.823 ms	0 - 395 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA7255P	173.118 ms	1 - 371 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA8295P	64.049 ms	0 - 306 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Elite Mobile	23.823 ms	0 - 395 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	19.433 ms	5 - 394 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	32.51 ms	0 - 500 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	71.262 ms	0 - 415 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® Dragonwing™ QCS8275	172.546 ms	0 - 394 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	44.215 ms	0 - 3 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® SA8775P	57.477 ms	0 - 393 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® SA8650P	57.477 ms	0 - 393 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® SA8255P	57.477 ms	0 - 393 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® QCS8450	71.262 ms	0 - 415 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	64.605 ms	0 - 89 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	23.787 ms	0 - 404 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® SA7255P	172.546 ms	0 - 394 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® SA8295P	63.627 ms	0 - 326 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Snapdragon® 8 Elite Mobile	23.787 ms	0 - 404 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	19.191 ms	0 - 414 MB	NPU

License

• The license for the original implementation of DETR-ResNet50-DC5 can be found here.

References

• End-to-End Object Detection with Transformers
• 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.