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 2.42, ONNX Runtime 1.25.0	Download
ONNX	w8a16_mixed_fp16	Universal	QAIRT 2.42, ONNX Runtime 1.25.0	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	w8a16_mixed_fp16	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:

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

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
DETR-ResNet50	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	7.32 ms	5 - 294 MB	NPU
DETR-ResNet50	ONNX	float	Snapdragon® X2 Elite	8.344 ms	207 - 207 MB	NPU
DETR-ResNet50	ONNX	float	Snapdragon® X Elite	17.71 ms	144 - 144 MB	NPU
DETR-ResNet50	ONNX	float	Snapdragon® 8 Gen 3 Mobile	13.408 ms	5 - 402 MB	NPU
DETR-ResNet50	ONNX	float	Qualcomm® QCS8550 (Proxy)	17.727 ms	0 - 103 MB	NPU
DETR-ResNet50	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	9.775 ms	1 - 288 MB	NPU
DETR-ResNet50	ONNX	float	Qualcomm® QCS9075	27.791 ms	5 - 50 MB	NPU
DETR-ResNet50	ONNX	float	Qualcomm® QCS8750	9.775 ms	1 - 288 MB	NPU
DETR-ResNet50	ONNX	float	Qualcomm® QCS7181	17.71 ms	144 - 144 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	22.727 ms	15 - 431 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	39.25 ms	17 - 499 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Qualcomm® QCS8550 (Proxy)	54.189 ms	14 - 113 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Qualcomm® QCS9075	86.692 ms	16 - 63 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite For Galaxy Mobile	31.175 ms	15 - 397 MB	NPU
DETR-ResNet50	ONNX	w8a16_mixed_fp16	Qualcomm® QCS8750	31.175 ms	15 - 397 MB	NPU
DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	7.816 ms	5 - 288 MB	NPU
DETR-ResNet50	QNN_DLC	float	Snapdragon® X2 Elite	9.048 ms	5 - 5 MB	NPU
DETR-ResNet50	QNN_DLC	float	Snapdragon® X Elite	19.506 ms	5 - 5 MB	NPU
DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	13.604 ms	0 - 351 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® QCS8275	86.103 ms	1 - 268 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	19.126 ms	5 - 10 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8775P	26.877 ms	1 - 300 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8650P	26.877 ms	1 - 300 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8255P	26.877 ms	1 - 300 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	42.183 ms	4 - 301 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® SA7255P	86.103 ms	1 - 268 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8295P	31.168 ms	0 - 224 MB	NPU
DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	10.474 ms	0 - 273 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® QCS9075	30.946 ms	5 - 11 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® QCS8750	10.474 ms	0 - 273 MB	NPU
DETR-ResNet50	QNN_DLC	float	Qualcomm® QCS7181	19.506 ms	5 - 5 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	16.939 ms	2 - 396 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® X2 Elite	18.286 ms	2 - 2 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® X Elite	32.7 ms	2 - 2 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	24.797 ms	2 - 444 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® QCS8275	94.224 ms	2 - 350 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® QCS8550 (Proxy)	33.716 ms	2 - 343 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8775P	38.482 ms	1 - 373 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8650P	38.482 ms	1 - 373 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8255P	38.482 ms	1 - 373 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® QCS9075	63.294 ms	2 - 6 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Elite For Galaxy Mobile	19.399 ms	2 - 360 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA7255P	94.224 ms	2 - 350 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® QCS8750	19.399 ms	2 - 360 MB	NPU
DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® QCS7181	32.7 ms	2 - 2 MB	NPU
DETR-ResNet50	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	7.959 ms	0 - 309 MB	NPU
DETR-ResNet50	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	13.68 ms	0 - 387 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® QCS8275	86.018 ms	0 - 293 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® QCS8550 (Proxy)	19.268 ms	0 - 3 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® SA8775P	26.924 ms	0 - 324 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® SA8650P	26.924 ms	0 - 324 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® SA8255P	26.924 ms	0 - 324 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® QCS8450 (Proxy)	41.922 ms	0 - 330 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® SA7255P	86.018 ms	0 - 293 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® SA8295P	31.1 ms	0 - 245 MB	NPU
DETR-ResNet50	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	10.501 ms	0 - 306 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® QCS9075	30.275 ms	0 - 88 MB	NPU
DETR-ResNet50	TFLITE	float	Qualcomm® QCS8750	10.501 ms	0 - 306 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.