Yolo-v3: Optimized for Qualcomm Devices

YoloV3 is a machine learning model that predicts bounding boxes and classes of objects in an image.

This is based on the implementation of Yolo-v3 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-v3 on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.object_detection

Model Stats:

• Model checkpoint: YoloV3 Tiny
• Input resolution: 416p (416x416)
• Number of parameters: 11.5M
• Model size (float): 43.9 MB
• Model size (w8a16): 16.9 MB

Performance Summary

Model	Runtime	Precision	Chipset	Inference Time (ms)	Peak Memory Range (MB)	Primary Compute Unit
Yolo-v3	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	1.855 ms	1 - 153 MB	NPU
Yolo-v3	ONNX	float	Snapdragon® 8 Elite Mobile	2.528 ms	0 - 142 MB	NPU
Yolo-v3	ONNX	float	Snapdragon® X2 Elite	2.419 ms	20 - 20 MB	NPU
Yolo-v3	ONNX	float	Snapdragon® X Elite	4.209 ms	19 - 19 MB	NPU
Yolo-v3	ONNX	float	Snapdragon® X Elite	4.209 ms	19 - 19 MB	NPU
Yolo-v3	ONNX	float	Snapdragon® 8 Gen 3 Mobile	3.174 ms	3 - 174 MB	NPU
Yolo-v3	ONNX	float	Qualcomm® QCS8550 (Proxy)	4.045 ms	0 - 23 MB	NPU
Yolo-v3	ONNX	float	Qualcomm® QCS9075	6.243 ms	5 - 7 MB	NPU
Yolo-v3	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	2.528 ms	0 - 142 MB	NPU
Yolo-v3	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	1.392 ms	0 - 55 MB	NPU
Yolo-v3	ONNX	w8a16	Snapdragon® 8 Elite Mobile	1.878 ms	0 - 48 MB	NPU
Yolo-v3	ONNX	w8a16	Snapdragon® X2 Elite	1.623 ms	10 - 10 MB	NPU
Yolo-v3	ONNX	w8a16	Snapdragon® X Elite	4.005 ms	10 - 10 MB	NPU
Yolo-v3	ONNX	w8a16	Snapdragon® X Elite	4.005 ms	10 - 10 MB	NPU
Yolo-v3	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	2.529 ms	0 - 83 MB	NPU
Yolo-v3	ONNX	w8a16	Qualcomm® QCS6490	527.172 ms	212 - 217 MB	CPU
Yolo-v3	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	3.683 ms	0 - 11 MB	NPU
Yolo-v3	ONNX	w8a16	Qualcomm® QCS9075	3.981 ms	2 - 5 MB	NPU
Yolo-v3	ONNX	w8a16	Qualcomm® QCM6690	263.445 ms	212 - 219 MB	CPU
Yolo-v3	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	1.878 ms	0 - 48 MB	NPU
Yolo-v3	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	237.619 ms	215 - 223 MB	CPU
Yolo-v3	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	237.619 ms	215 - 223 MB	CPU
Yolo-v3	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	1.664 ms	5 - 153 MB	NPU
Yolo-v3	QNN_DLC	float	Snapdragon® 8 Elite Mobile	2.152 ms	5 - 148 MB	NPU
Yolo-v3	QNN_DLC	float	Snapdragon® X2 Elite	2.456 ms	5 - 5 MB	NPU
Yolo-v3	QNN_DLC	float	Snapdragon® X Elite	3.7 ms	5 - 5 MB	NPU
Yolo-v3	QNN_DLC	float	Snapdragon® X Elite	3.7 ms	5 - 5 MB	NPU
Yolo-v3	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	2.746 ms	0 - 162 MB	NPU
Yolo-v3	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	20.095 ms	1 - 143 MB	NPU
Yolo-v3	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	3.393 ms	2 - 4 MB	NPU
Yolo-v3	QNN_DLC	float	Qualcomm® QCS9075	5.799 ms	5 - 11 MB	NPU
Yolo-v3	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	10.035 ms	4 - 173 MB	NPU
Yolo-v3	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	2.152 ms	5 - 148 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	1.393 ms	2 - 52 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® 8 Elite Mobile	1.886 ms	0 - 51 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® X2 Elite	1.756 ms	2 - 2 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® X Elite	4.044 ms	2 - 2 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® X Elite	4.044 ms	2 - 2 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	2.57 ms	2 - 75 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCS6490	11.54 ms	2 - 6 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCS8275 (Proxy)	9.173 ms	1 - 46 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	3.666 ms	2 - 4 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCS9075	3.937 ms	1 - 5 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCM6690	37.371 ms	2 - 174 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCS8450 (Proxy)	4.464 ms	2 - 77 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	1.886 ms	0 - 51 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	4.674 ms	2 - 173 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	4.674 ms	2 - 173 MB	NPU
Yolo-v3	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	1.533 ms	0 - 148 MB	NPU
Yolo-v3	TFLITE	float	Snapdragon® 8 Elite Mobile	2.113 ms	0 - 150 MB	NPU
Yolo-v3	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	2.678 ms	0 - 182 MB	NPU
Yolo-v3	TFLITE	float	Qualcomm® QCS8275 (Proxy)	19.955 ms	0 - 145 MB	NPU
Yolo-v3	TFLITE	float	Qualcomm® QCS8550 (Proxy)	3.322 ms	0 - 16 MB	NPU
Yolo-v3	TFLITE	float	Qualcomm® QCS9075	5.579 ms	0 - 30 MB	NPU
Yolo-v3	TFLITE	float	Qualcomm® QCS8450 (Proxy)	9.948 ms	0 - 192 MB	NPU
Yolo-v3	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	2.113 ms	0 - 150 MB	NPU

License

• The license for the original implementation of Yolo-v3 can be found here.

References

• YOLOv3: An Incremental Improvement
• 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.