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 1.25.0	Download
ONNX	w8a16_mixed_int16	Universal	QAIRT 2.45, ONNX Runtime 1.25.0	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	19.947 ms	208 - 208 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® X Elite	44.751 ms	144 - 144 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® 8 Gen 3 Mobile	31.663 ms	5 - 480 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® 8 Gen 1 Mobile	68.279 ms	8 - 407 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS8550 (Proxy)	43.039 ms	0 - 96 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS8450	68.279 ms	8 - 407 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	18.356 ms	3 - 380 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS9075	63.025 ms	5 - 50 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS8750	23.038 ms	3 - 374 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS7181	44.751 ms	144 - 144 MB	NPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Snapdragon® 8 Gen 3 Mobile	153.846 ms	56 - 628 MB	NPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Snapdragon® 8 Elite Mobile	135.714 ms	52 - 504 MB	NPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Snapdragon® 8 Elite Gen 5 Mobile	122.85 ms	55 - 539 MB	NPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Qualcomm® QCS9075	191.055 ms	55 - 98 MB	NPU
DETR-ResNet50-DC5	ONNX	w8a16_mixed_int16	Qualcomm® QCS8750	135.714 ms	52 - 504 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® X2 Elite	21.226 ms	5 - 5 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® X Elite	46.624 ms	5 - 5 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	32.545 ms	5 - 466 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	71.545 ms	4 - 396 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8275	173.103 ms	0 - 371 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	44.657 ms	5 - 7 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8450	71.545 ms	4 - 396 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Elite Mobile	23.736 ms	5 - 399 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA8295P	64.078 ms	0 - 306 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	19.362 ms	5 - 394 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA7255P	173.103 ms	0 - 371 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS9075	67.566 ms	7 - 13 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8750	23.736 ms	5 - 399 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS7181	46.624 ms	5 - 5 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	32.417 ms	0 - 499 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	71.358 ms	0 - 416 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® QCS8275	172.577 ms	1 - 393 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® QCS8550 (Proxy)	44.231 ms	0 - 3 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® SA8775P	1210.12 ms	0 - 11 MB	CPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® SA8650P	1210.12 ms	0 - 11 MB	CPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® SA8255P	1210.12 ms	0 - 11 MB	CPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® QCS8450	71.358 ms	0 - 416 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Snapdragon® 8 Elite Mobile	23.749 ms	0 - 406 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® SA8295P	63.594 ms	0 - 326 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	19.166 ms	1 - 416 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® SA7255P	172.577 ms	1 - 393 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® QCS9075	67.604 ms	0 - 89 MB	NPU
DETR-ResNet50-DC5	TFLITE	float	Qualcomm® QCS8750	23.749 ms	0 - 406 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.