DETR-ResNet50: 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 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
ONNX	w8a16_mixed_fp16	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 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 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.object_detection

Model Stats:

• Input resolution: 480x480
• Model checkpoint: ResNet50
• Model size (float): 158 MB
• Number of parameters: 41.4M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
DETR-ResNet50	ONNX	float	Snapdragon® X2 Elite	8.242 ms	5 - 5 MB	NPU
DETR-ResNet50	ONNX	float	Snapdragon® X Elite	17.971 ms	78 - 78 MB	NPU
DETR-ResNet50	ONNX	float	Snapdragon® 8 Gen 3 Mobile	13.419 ms	5 - 374 MB	NPU
DETR-ResNet50	ONNX	float	Snapdragon® 8 Gen 1 Mobile	36.007 ms	1 - 318 MB	NPU
DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	29.042 ms	5 - 13 MB	NPU
DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	17.775 ms	0 - 94 MB	NPU
DETR-ResNet50	ONNX	float	Qualcomm® QCS8450	36.007 ms	1 - 318 MB	NPU
DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	27.57 ms	4 - 12 MB	NPU
DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	17.971 ms	78 - 78 MB	NPU
DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ Q-8750	9.975 ms	3 - 271 MB	NPU
DETR-ResNet50	ONNX	float	Snapdragon® 8 Elite Mobile	9.975 ms	3 - 271 MB	NPU
DETR-ResNet50	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	7.436 ms	2 - 317 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	39.415 ms	6 - 351 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Qualcomm® Dragonwing™ IQ-8275	68.232 ms	16 - 19 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	50.417 ms	16 - 21 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Qualcomm® QCS8450	97.068 ms	7 - 466 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Qualcomm® Dragonwing™ IQ-9075	51.175 ms	16 - 19 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Qualcomm® Dragonwing™ Q-8750	33.662 ms	15 - 275 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite Mobile	33.662 ms	15 - 275 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	28.229 ms	15 - 283 MB	NPU
DETR-ResNet50	QNN_DLC	float	Snapdragon® X2 Elite	8.917 ms	5 - 5 MB	NPU
DETR-ResNet50	QNN_DLC	float	Snapdragon® X Elite	19.561 ms	5 - 5 MB	NPU
DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	13.7 ms	2 - 357 MB	NPU
DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	41.417 ms	5 - 303 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	27.712 ms	5 - 12 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	19.181 ms	5 - 7 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8775P	26.853 ms	1 - 300 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8650P	26.853 ms	1 - 300 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8255P	26.853 ms	1 - 300 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® QCS8450	41.417 ms	5 - 303 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	27.178 ms	7 - 13 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	19.561 ms	5 - 5 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	10.448 ms	0 - 273 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® SA7255P	86.011 ms	1 - 268 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8295P	31.185 ms	0 - 224 MB	NPU
DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Elite Mobile	10.448 ms	0 - 273 MB	NPU
DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	7.898 ms	5 - 290 MB	NPU
DETR-ResNet50	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	13.771 ms	0 - 390 MB	NPU
DETR-ResNet50	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	41.717 ms	0 - 327 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	27.739 ms	0 - 89 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	19.235 ms	0 - 2 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® SA8775P	26.917 ms	0 - 323 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® SA8650P	26.917 ms	0 - 323 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® SA8255P	26.917 ms	0 - 323 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® QCS8450	41.717 ms	0 - 327 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	27.256 ms	0 - 88 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	10.497 ms	0 - 302 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® SA7255P	86.011 ms	0 - 293 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® SA8295P	31.092 ms	0 - 247 MB	NPU
DETR-ResNet50	TFLITE	float	Snapdragon® 8 Elite Mobile	10.497 ms	0 - 302 MB	NPU
DETR-ResNet50	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	8.047 ms	0 - 313 MB	NPU

License

• The license for the original implementation of DETR-ResNet50 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.