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 2.37, ONNX Runtime 1.23.0	Download
QNN_DLC	float	Universal	QAIRT 2.42	Download
TFLITE	float	Universal	QAIRT 2.42, TFLite 2.17.0	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® X Elite	43.461 ms	79 - 79 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® 8 Gen 3 Mobile	30.834 ms	5 - 595 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS8550 (Proxy)	42.75 ms	0 - 94 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Qualcomm® QCS9075	60.037 ms	4 - 12 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	22.115 ms	0 - 364 MB	NPU
DETR-ResNet50-DC5	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	18.629 ms	1 - 399 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® X Elite	47.518 ms	5 - 5 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	33.857 ms	4 - 633 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	171.882 ms	2 - 480 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	46.504 ms	5 - 7 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA8775P	60.117 ms	1 - 482 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS9075	69.403 ms	5 - 11 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	74.472 ms	2 - 447 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA7255P	171.882 ms	2 - 480 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Qualcomm® SA8295P	63.643 ms	0 - 331 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	24.139 ms	0 - 496 MB	NPU
DETR-ResNet50-DC5	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	19.825 ms	5 - 476 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.