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:

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

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Yolo-R	ONNX	float	Snapdragon® X2 Elite	26.281 ms	208 - 208 MB	NPU
Yolo-R	ONNX	float	Snapdragon® X Elite	48.569 ms	144 - 144 MB	NPU
Yolo-R	ONNX	float	Snapdragon® 8 Gen 3 Mobile	37.375 ms	6 - 325 MB	NPU
Yolo-R	ONNX	float	Snapdragon® 8 Gen 1 Mobile	62.973 ms	6 - 374 MB	NPU
Yolo-R	ONNX	float	Qualcomm® QCS8550 (Proxy)	48.26 ms	0 - 80 MB	NPU
Yolo-R	ONNX	float	Qualcomm® QCS8450	62.973 ms	6 - 374 MB	NPU
Yolo-R	ONNX	float	Snapdragon® 8 Elite Mobile	27.253 ms	1 - 231 MB	NPU
Yolo-R	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	26.587 ms	2 - 292 MB	NPU
Yolo-R	ONNX	float	Qualcomm® QCS9075	54.75 ms	5 - 50 MB	NPU
Yolo-R	ONNX	float	Qualcomm® QCS8750	27.253 ms	1 - 231 MB	NPU
Yolo-R	ONNX	float	Qualcomm® QCS7181	48.569 ms	144 - 144 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® X2 Elite	18.191 ms	179 - 179 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® X Elite	30.065 ms	148 - 148 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	20.838 ms	0 - 448 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® 8 Gen 1 Mobile	38.253 ms	4 - 452 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	28.747 ms	0 - 53 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® QCS8450	38.253 ms	4 - 452 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® QCS9075	29.573 ms	2 - 48 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	17.801 ms	1 - 421 MB	NPU
Yolo-R	ONNX	w8a16	Snapdragon® 8 Elite Mobile	17.071 ms	1 - 365 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® QCS8750	17.071 ms	1 - 365 MB	NPU
Yolo-R	ONNX	w8a16	Qualcomm® QCS7181	30.065 ms	148 - 148 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® X2 Elite	8.757 ms	2 - 2 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® X Elite	20.361 ms	2 - 2 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	13.118 ms	2 - 358 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® 8 Gen 1 Mobile	27.847 ms	2 - 359 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® QCS6490	74.958 ms	2 - 7 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® QCS8275	39.545 ms	1 - 293 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	19.601 ms	2 - 5 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® QCS8450	27.847 ms	2 - 359 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® QCS9075	20.366 ms	2 - 7 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	7.717 ms	2 - 311 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	9.815 ms	2 - 305 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® SA8295P	25.044 ms	0 - 294 MB	NPU
Yolo-R	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	25.938 ms	2 - 312 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® SA7255P	39.545 ms	1 - 293 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® QCM6690	222.028 ms	2 - 398 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® QCS7790	25.938 ms	2 - 312 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® QCS8750	9.815 ms	2 - 305 MB	NPU
Yolo-R	QNN_DLC	w8a16	Qualcomm® QCS7181	20.361 ms	2 - 2 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.