DETR-ResNet101-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-ResNet101-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-ResNet101-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-ResNet101-DC5 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.object_detection

Model Stats:

• Model checkpoint: ResNet101-DC5
• Input resolution: 480x480
• Model size (float): 232 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
DETR-ResNet101-DC5	ONNX	float	Snapdragon® X2 Elite	23.617 ms	5 - 5 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Snapdragon® X Elite	51.355 ms	115 - 115 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Snapdragon® 8 Gen 3 Mobile	36.478 ms	5 - 575 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Snapdragon® 8 Gen 1 Mobile	84.27 ms	0 - 478 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	49.721 ms	5 - 7 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Qualcomm® QCS8450	84.27 ms	0 - 478 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	68.853 ms	5 - 12 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	51.355 ms	115 - 115 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Qualcomm® Dragonwing™ Q-8750	27.287 ms	2 - 436 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Snapdragon® 8 Elite Mobile	27.287 ms	2 - 436 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	21.773 ms	3 - 445 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Snapdragon® X2 Elite	24.921 ms	5 - 5 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Snapdragon® X Elite	53.27 ms	5 - 5 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	37.538 ms	5 - 556 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	108.927 ms	5 - 469 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8275	220.397 ms	2 - 427 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	51.38 ms	5 - 7 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® SA8775P	69.887 ms	1 - 427 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® SA8650P	69.887 ms	1 - 427 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® SA8255P	69.887 ms	1 - 427 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® QCS8450	108.927 ms	5 - 469 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	69.638 ms	7 - 13 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	53.27 ms	5 - 5 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	28.135 ms	0 - 450 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® SA7255P	220.397 ms	2 - 427 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® SA8295P	86.058 ms	1 - 342 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Snapdragon® 8 Elite Mobile	28.135 ms	0 - 450 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	22.662 ms	5 - 450 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	37.198 ms	0 - 599 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	84.692 ms	1 - 478 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Qualcomm® Dragonwing™ QCS8275	219.41 ms	0 - 471 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	50.75 ms	0 - 4 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Qualcomm® SA8775P	69.558 ms	0 - 470 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Qualcomm® SA8650P	69.558 ms	0 - 470 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Qualcomm® SA8255P	69.558 ms	0 - 470 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Qualcomm® QCS8450	84.692 ms	1 - 478 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	78.078 ms	0 - 126 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	27.746 ms	0 - 476 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Qualcomm® SA7255P	219.41 ms	0 - 471 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Qualcomm® SA8295P	76.615 ms	0 - 358 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Snapdragon® 8 Elite Mobile	27.746 ms	0 - 476 MB	NPU
DETR-ResNet101-DC5	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	22.313 ms	0 - 494 MB	NPU

License

• The license for the original implementation of DETR-ResNet101-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.