Yolo-R: Optimized for Qualcomm Devices

YoloR is a machine learning model that predicts bounding boxes and classes of objects in an image.

This is based on the implementation of Yolo-R 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

Due to licensing restrictions, we cannot distribute pre-exported model assets for this model. 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

See our repository for Yolo-R on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.object_detection

Model Stats:

• Input resolution: 640x640
• Model checkpoint: yolor_p6
• Model size (float): 17.9 MB
• Number of parameters: 4.68M

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Yolo-R	ONNX	float	Snapdragon® X2 Elite	27.641 ms	5 - 5 MB	NPU
Yolo-R	ONNX	float	Snapdragon® X Elite	45.687 ms	71 - 71 MB	NPU
Yolo-R	ONNX	float	Snapdragon® 8 Gen 3 Mobile	35.369 ms	0 - 250 MB	NPU
Yolo-R	ONNX	float	Snapdragon® 8 Gen 1 Mobile	72.043 ms	1 - 311 MB	NPU
Yolo-R	ONNX	float	Qualcomm® Dragonwing™ IQ-8275	53.883 ms	5 - 13 MB	NPU
Yolo-R	ONNX	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	45.331 ms	5 - 8 MB	NPU
Yolo-R	ONNX	float	Qualcomm® QCS8450	72.043 ms	1 - 311 MB	NPU
Yolo-R	ONNX	float	Qualcomm® Dragonwing™ IQ-9075	52.819 ms	5 - 12 MB	NPU
Yolo-R	ONNX	float	Qualcomm® Dragonwing™ IQ-X7181	45.687 ms	71 - 71 MB	NPU
Yolo-R	ONNX	float	Qualcomm® Dragonwing™ Q-8750	31.397 ms	3 - 199 MB	NPU
Yolo-R	ONNX	float	Snapdragon® 8 Elite Mobile	31.397 ms	3 - 199 MB	NPU
Yolo-R	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	27.832 ms	3 - 203 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® X2 Elite	15.295 ms	2 - 2 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® X Elite	23.627 ms	37 - 37 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	16.002 ms	0 - 397 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	32.197 ms	3 - 405 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-8275	21.457 ms	1 - 7 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	22.716 ms	2 - 5 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® QCS8450	32.197 ms	3 - 405 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-9075	23.651 ms	1 - 6 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® Dragonwing™ IQ-X7181	23.627 ms	37 - 37 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® Dragonwing™ Q-7790	47.813 ms	3 - 320 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® Dragonwing™ Q-8750	13.536 ms	1 - 334 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® 8 Elite Mobile	13.536 ms	1 - 334 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	14.638 ms	1 - 343 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	47.813 ms	3 - 320 MB	NPU
Yolo-R	QNN_DLC	float	Snapdragon® X2 Elite	18.273 ms	5 - 5 MB	NPU
Yolo-R	QNN_DLC	float	Snapdragon® X Elite	51.739 ms	5 - 5 MB	NPU
Yolo-R	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	41.087 ms	5 - 259 MB	NPU
Yolo-R	QNN_DLC	float	Snapdragon® 8 Gen 1 Mobile	65.212 ms	5 - 312 MB	NPU
Yolo-R	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	58.465 ms	5 - 12 MB	NPU
Yolo-R	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-8275	124.423 ms	2 - 199 MB	NPU
Yolo-R	QNN_DLC	float	Qualcomm® Dragonwing™ QCS8550 (Proxy)	55.737 ms	5 - 6 MB	NPU
Yolo-R	QNN_DLC	float	Qualcomm® SA8775P	61.398 ms	1 - 198 MB	NPU
Yolo-R	QNN_DLC	float	Qualcomm® SA8650P	61.398 ms	1 - 198 MB	NPU
Yolo-R	QNN_DLC	float	Qualcomm® SA8255P	61.398 ms	1 - 198 MB	NPU
Yolo-R	QNN_DLC	float	Qualcomm® QCS8450	65.212 ms	5 - 312 MB	NPU
Yolo-R	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-9075	61.393 ms	5 - 11 MB	NPU
Yolo-R	QNN_DLC	float	Qualcomm® Dragonwing™ IQ-X7181	51.739 ms	5 - 5 MB	NPU
Yolo-R	QNN_DLC	float	Qualcomm® Dragonwing™ Q-8750	22.644 ms	0 - 195 MB	NPU
Yolo-R	QNN_DLC	float	Qualcomm® SA7255P	124.423 ms	2 - 199 MB	NPU
Yolo-R	QNN_DLC	float	Snapdragon® 8 Elite Mobile	22.644 ms	0 - 195 MB	NPU
Yolo-R	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	19.113 ms	5 - 212 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® X2 Elite	7.088 ms	2 - 2 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® X Elite	16.257 ms	2 - 2 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	10.01 ms	2 - 381 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® 8 Gen 1 Mobile	25.631 ms	2 - 382 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS6490	64.981 ms	4 - 8 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	13.964 ms	2 - 7 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-8275	32.588 ms	2 - 311 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® Dragonwing™ QCS8550 (Proxy)	15.257 ms	2 - 5 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® SA8775P	15.473 ms	1 - 310 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® SA8650P	15.473 ms	1 - 310 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® SA8255P	15.473 ms	1 - 310 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® QCS8450	25.631 ms	2 - 382 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-9075	15.845 ms	2 - 7 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® Dragonwing™ IQ-X7181	16.257 ms	2 - 2 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-6690	164.995 ms	3 - 351 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-7790	21.571 ms	2 - 326 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® Dragonwing™ Q-8750	7.343 ms	2 - 322 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® SA7255P	32.588 ms	2 - 311 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® SA8295P	20.794 ms	0 - 312 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	7.343 ms	2 - 322 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	5.864 ms	2 - 327 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	21.571 ms	2 - 326 MB	NPU

License

• The license for the original implementation of Yolo-R can be found here.

References

• You Only Learn One Representation: Unified Network for Multiple Tasks
• 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.