Conditional-DETR-ResNet50: 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 Conditional-DETR-ResNet50 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_fp16	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 Conditional-DETR-ResNet50 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 Conditional-DETR-ResNet50 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.object_detection

Model Stats:

• Model checkpoint: ResNet50
• Input resolution: 480x480
• Number of parameters: 43.6M
• Model size (float): 166 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® X2 Elite	9.266 ms	5 - 5 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® X Elite	20.313 ms	82 - 82 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® 8 Gen 3 Mobile	15.009 ms	1 - 453 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® 8 Gen 1 Mobile	39.083 ms	5 - 374 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	20.231 ms	0 - 94 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Qualcomm® QCS8450	39.083 ms	5 - 374 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	29.105 ms	5 - 12 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	8.214 ms	1 - 319 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® 8 Elite Mobile	11.122 ms	3 - 315 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ Q-8750	11.122 ms	3 - 315 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	20.313 ms	82 - 82 MB	NPU
Conditional-DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	61.017 ms	20 - 497 MB	NPU
Conditional-DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite Mobile	49.825 ms	21 - 399 MB	NPU
Conditional-DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	40.495 ms	23 - 436 MB	NPU
Conditional-DETR-ResNet50	ONNX	w8a16_mixed_fp16	Qualcomm® Dragonwing™ Q-8750	49.825 ms	21 - 399 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® X2 Elite	9.996 ms	5 - 5 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® X Elite	23.412 ms	5 - 5 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	17.089 ms	5 - 401 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	44.609 ms	5 - 349 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® QCS8275	97.953 ms	1 - 309 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	23.177 ms	5 - 8 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8775P	32.236 ms	1 - 308 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8650P	32.236 ms	1 - 308 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8255P	32.236 ms	1 - 308 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® QCS8450	44.609 ms	5 - 349 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	33.223 ms	5 - 11 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	8.948 ms	5 - 338 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® SA7255P	97.953 ms	1 - 309 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Elite Mobile	11.615 ms	0 - 316 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8295P	34.246 ms	0 - 260 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	11.615 ms	0 - 316 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	23.412 ms	5 - 5 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® X2 Elite	18.268 ms	2 - 2 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® X Elite	36.537 ms	2 - 2 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	26.754 ms	2 - 505 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® QCS8275	102.505 ms	1 - 403 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	36.449 ms	2 - 6 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8775P	41.802 ms	1 - 424 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8650P	41.802 ms	1 - 424 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8255P	41.802 ms	1 - 424 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Elite Mobile	19.654 ms	0 - 411 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	16.478 ms	2 - 456 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® Dragonwing™ IQ-9075	42.271 ms	2 - 6 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA7255P	102.505 ms	1 - 403 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® Dragonwing™ Q-8750	19.654 ms	0 - 411 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® Dragonwing™ IQ-X7181	36.537 ms	2 - 2 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	15.811 ms	0 - 437 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	45.356 ms	0 - 374 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® QCS8275	93.314 ms	0 - 338 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	22.041 ms	0 - 3 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® SA8775P	29.982 ms	0 - 368 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® SA8650P	29.982 ms	0 - 368 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® SA8255P	29.982 ms	0 - 368 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® QCS8450	45.356 ms	0 - 374 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	33.0 ms	0 - 93 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	9.299 ms	0 - 354 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® SA7255P	93.314 ms	0 - 338 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Snapdragon® 8 Elite Mobile	11.898 ms	0 - 340 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® SA8295P	34.672 ms	0 - 288 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	11.898 ms	0 - 340 MB	NPU

License

• The license for the original implementation of Conditional-DETR-ResNet50 can be found here.

References

• Conditional {DETR} for Fast Training Convergence
• 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.