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 1.27.1	Download
ONNX	w8a16_mixed_fp16	Universal	QAIRT 2.45, ONNX Runtime 1.27.1	Download
QNN_DLC	float	Universal	QAIRT 2.45	Download
QNN_DLC	w8a16_mixed_fp16	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:

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

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® X2 Elite	9.27 ms	5 - 5 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® X Elite	20.35 ms	82 - 82 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® 8 Gen 3 Mobile	15.097 ms	5 - 458 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® 8 Gen 1 Mobile	39.099 ms	1 - 369 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	31.378 ms	5 - 13 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	20.21 ms	0 - 94 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Qualcomm® QCS8450	39.099 ms	1 - 369 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	29.926 ms	5 - 12 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	20.35 ms	82 - 82 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Qualcomm® Dragonwing™ Q-8750	11.189 ms	1 - 318 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® 8 Elite Mobile	11.189 ms	1 - 318 MB	NPU
Conditional-DETR-ResNet50	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	8.305 ms	1 - 323 MB	NPU
Conditional-DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	52.478 ms	19 - 405 MB	NPU
Conditional-DETR-ResNet50	ONNX	w8a16_mixed_fp16	Qualcomm® Dragonwing™ Q-8750	46.115 ms	23 - 296 MB	NPU
Conditional-DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite Mobile	46.115 ms	23 - 296 MB	NPU
Conditional-DETR-ResNet50	ONNX	w8a16_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	39.232 ms	22 - 311 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® X2 Elite	10.112 ms	5 - 5 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® X Elite	23.438 ms	5 - 5 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	17.11 ms	5 - 403 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	44.662 ms	4 - 347 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	33.265 ms	5 - 12 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	98.249 ms	1 - 309 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	23.041 ms	5 - 7 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8775P	32.198 ms	1 - 306 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8650P	32.198 ms	1 - 306 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8255P	32.198 ms	1 - 306 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® QCS8450	44.662 ms	4 - 347 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	32.756 ms	7 - 13 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	23.438 ms	5 - 5 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	11.584 ms	0 - 317 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® SA7255P	98.249 ms	1 - 309 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Qualcomm® SA8295P	34.07 ms	0 - 262 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Elite Mobile	11.584 ms	0 - 317 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	8.963 ms	5 - 338 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® X2 Elite	16.452 ms	2 - 2 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® X Elite	30.561 ms	2 - 2 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Gen 3 Mobile	22.475 ms	2 - 475 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® Dragonwing™ IQ-8275	35.661 ms	0 - 5 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® Dragonwing™ IQ-8275	90.15 ms	1 - 370 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	30.292 ms	2 - 5 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8775P	35.852 ms	1 - 370 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8650P	35.852 ms	1 - 370 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA8255P	35.852 ms	1 - 370 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® Dragonwing™ IQ-9075	36.296 ms	1 - 5 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® Dragonwing™ IQ-X7181	30.561 ms	2 - 2 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® Dragonwing™ Q-8750	17.486 ms	2 - 374 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Qualcomm® SA7255P	90.15 ms	1 - 370 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Elite Mobile	17.486 ms	2 - 374 MB	NPU
Conditional-DETR-ResNet50	QNN_DLC	w8a16_mixed_fp16	Snapdragon® 8 Elite Gen 5 Mobile	15.085 ms	2 - 393 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	15.928 ms	0 - 434 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Snapdragon® 8 Gen 1 Mobile	46.588 ms	2 - 374 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	31.224 ms	0 - 94 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ IQ-8275	94.004 ms	0 - 334 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	22.42 ms	0 - 2 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® SA8775P	30.265 ms	0 - 367 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® SA8650P	30.265 ms	0 - 367 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® SA8255P	30.265 ms	0 - 367 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® QCS8450	46.588 ms	2 - 374 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ IQ-9075	30.228 ms	0 - 93 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® Dragonwing™ Q-8750	12.467 ms	0 - 346 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® SA7255P	94.004 ms	0 - 334 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Qualcomm® SA8295P	35.196 ms	0 - 289 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Snapdragon® 8 Elite Mobile	12.467 ms	0 - 346 MB	NPU
Conditional-DETR-ResNet50	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	9.564 ms	0 - 355 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.