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 2.42, ONNX Runtime 1.24.3	Download
QNN_DLC	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® 8 Elite Gen 5 Mobile	23.439 ms	5 - 529 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Snapdragon® X2 Elite	25.636 ms	116 - 116 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Snapdragon® X Elite	54.309 ms	115 - 115 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Snapdragon® 8 Gen 3 Mobile	38.872 ms	5 - 658 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Qualcomm® QCS8550 (Proxy)	52.399 ms	0 - 128 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Qualcomm® QCS9075	80.87 ms	4 - 12 MB	NPU
DETR-ResNet101-DC5	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	32.361 ms	2 - 516 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	28.133 ms	5 - 589 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Snapdragon® X2 Elite	28.847 ms	5 - 5 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Snapdragon® X Elite	62.181 ms	5 - 5 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	44.93 ms	4 - 763 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	232.936 ms	2 - 557 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	61.399 ms	5 - 8 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® SA8775P	79.765 ms	1 - 555 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® QCS9075	92.006 ms	7 - 13 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	108.687 ms	4 - 568 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® SA7255P	232.936 ms	2 - 557 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Qualcomm® SA8295P	94.178 ms	0 - 417 MB	NPU
DETR-ResNet101-DC5	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	31.742 ms	5 - 579 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.