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 2.42, ONNX Runtime 1.25.0	Download
QNN_DLC	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® 8 Elite Gen 5 Mobile	18.355 ms	5 - 461 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® X2 Elite	19.723 ms	177 - 177 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® X Elite	43.05 ms	171 - 171 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® 8 Gen 3 Mobile	31.48 ms	5 - 639 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS8550 (Proxy)	42.305 ms	0 - 727 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	21.741 ms	2 - 433 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS9075	62.392 ms	5 - 50 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS8750	21.741 ms	2 - 433 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS7181	43.05 ms	171 - 171 MB	NPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Snapdragon® 8 Elite Gen 5 Mobile	113.057 ms	44 - 543 MB	NPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Snapdragon® 8 Gen 3 Mobile	147.845 ms	55 - 742 MB	NPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Qualcomm® QCS6490	4472.245 ms	338 - 347 MB	CPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Snapdragon® 8 Elite For Galaxy Mobile	136.876 ms	52 - 528 MB	NPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Qualcomm® QCM6690	2143.696 ms	323 - 341 MB	CPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Snapdragon® 7 Gen 4 Mobile	2060.307 ms	335 - 354 MB	CPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Qualcomm® QCS7790	2060.307 ms	335 - 354 MB	CPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Qualcomm® QCS8750	136.876 ms	52 - 528 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	19.478 ms	5 - 391 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® X2 Elite	21.224 ms	5 - 5 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® X Elite	46.632 ms	5 - 5 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	32.536 ms	1 - 463 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8275	173.06 ms	5 - 376 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	44.502 ms	5 - 7 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA8775P	57.754 ms	1 - 371 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA8650P	57.754 ms	1 - 371 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA8255P	57.754 ms	1 - 371 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	23.832 ms	0 - 395 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	71.875 ms	4 - 397 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA8295P	64.052 ms	1 - 307 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS9075	67.898 ms	5 - 11 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA7255P	173.06 ms	5 - 376 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8750	23.832 ms	0 - 395 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS7181	46.632 ms	5 - 5 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Snapdragon® 8 Elite Gen 5 Mobile	22.837 ms	2 - 701 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Snapdragon® X2 Elite	25.249 ms	2 - 2 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Snapdragon® X Elite	50.521 ms	2 - 2 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Snapdragon® 8 Gen 3 Mobile	37.865 ms	2 - 791 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Qualcomm® QCS8275	104.621 ms	2 - 651 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Qualcomm® QCS8550 (Proxy)	49.171 ms	3 - 5 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Qualcomm® SA8775P	49.581 ms	1 - 526 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Qualcomm® SA8650P	49.581 ms	1 - 526 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Qualcomm® SA8255P	49.581 ms	1 - 526 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Snapdragon® 8 Elite For Galaxy Mobile	33.605 ms	2 - 665 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Qualcomm® QCM6690	570.814 ms	2 - 663 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Snapdragon® 7 Gen 4 Mobile	71.679 ms	2 - 624 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Qualcomm® QCS9075	61.71 ms	2 - 6 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Qualcomm® SA7255P	104.621 ms	2 - 651 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Qualcomm® QCS7790	71.679 ms	2 - 624 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Qualcomm® QCS8750	33.605 ms	2 - 665 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	w8a16_mixed_int16	Qualcomm® QCS7181	50.521 ms	2 - 2 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.