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® X2 Elite	2.414 ms	20 - 20 MB	NPU
Yolo-v3	ONNX	float	Snapdragon® X Elite	4.202 ms	19 - 19 MB	NPU
Yolo-v3	ONNX	float	Snapdragon® 8 Gen 3 Mobile	3.228 ms	2 - 174 MB	NPU
Yolo-v3	ONNX	float	Qualcomm® QCS8550 (Proxy)	3.994 ms	0 - 21 MB	NPU
Yolo-v3	ONNX	float	Qualcomm® QCS9075	6.203 ms	5 - 7 MB	NPU
Yolo-v3	ONNX	float	Snapdragon® 8 Elite For Galaxy Mobile	2.525 ms	0 - 150 MB	NPU
Yolo-v3	ONNX	float	Snapdragon® 8 Elite Gen 5 Mobile	1.81 ms	1 - 150 MB	NPU
Yolo-v3	ONNX	w8a16	Snapdragon® X2 Elite	1.625 ms	10 - 10 MB	NPU
Yolo-v3	ONNX	w8a16	Snapdragon® X Elite	3.994 ms	10 - 10 MB	NPU
Yolo-v3	ONNX	w8a16	Snapdragon® 8 Gen 3 Mobile	2.524 ms	0 - 91 MB	NPU
Yolo-v3	ONNX	w8a16	Qualcomm® QCS6490	528.673 ms	213 - 218 MB	CPU
Yolo-v3	ONNX	w8a16	Qualcomm® QCS8550 (Proxy)	3.689 ms	0 - 12 MB	NPU
Yolo-v3	ONNX	w8a16	Qualcomm® QCS9075	4.017 ms	2 - 5 MB	NPU
Yolo-v3	ONNX	w8a16	Qualcomm® QCM6690	261.852 ms	125 - 131 MB	CPU
Yolo-v3	ONNX	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	1.88 ms	1 - 55 MB	NPU
Yolo-v3	ONNX	w8a16	Snapdragon® 7 Gen 4 Mobile	229.443 ms	200 - 208 MB	CPU
Yolo-v3	ONNX	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	1.391 ms	0 - 56 MB	NPU
Yolo-v3	QNN_DLC	float	Snapdragon® X2 Elite	2.402 ms	5 - 5 MB	NPU
Yolo-v3	QNN_DLC	float	Snapdragon® X Elite	3.659 ms	5 - 5 MB	NPU
Yolo-v3	QNN_DLC	float	Snapdragon® 8 Gen 3 Mobile	2.74 ms	5 - 169 MB	NPU
Yolo-v3	QNN_DLC	float	Qualcomm® QCS8275 (Proxy)	20.12 ms	0 - 143 MB	NPU
Yolo-v3	QNN_DLC	float	Qualcomm® QCS8550 (Proxy)	3.387 ms	4 - 5 MB	NPU
Yolo-v3	QNN_DLC	float	Qualcomm® QCS9075	5.758 ms	5 - 11 MB	NPU
Yolo-v3	QNN_DLC	float	Qualcomm® QCS8450 (Proxy)	10.037 ms	5 - 176 MB	NPU
Yolo-v3	QNN_DLC	float	Snapdragon® 8 Elite For Galaxy Mobile	2.16 ms	5 - 150 MB	NPU
Yolo-v3	QNN_DLC	float	Snapdragon® 8 Elite Gen 5 Mobile	1.732 ms	5 - 152 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® X2 Elite	1.846 ms	2 - 2 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® X Elite	4.049 ms	2 - 2 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® 8 Gen 3 Mobile	2.567 ms	2 - 81 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCS6490	11.789 ms	2 - 6 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCS8275 (Proxy)	9.245 ms	0 - 45 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCS8550 (Proxy)	3.714 ms	2 - 4 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCS9075	3.931 ms	2 - 6 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCM6690	35.199 ms	2 - 174 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Qualcomm® QCS8450 (Proxy)	4.378 ms	2 - 78 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® 8 Elite For Galaxy Mobile	1.897 ms	2 - 52 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® 7 Gen 4 Mobile	4.701 ms	2 - 174 MB	NPU
Yolo-v3	QNN_DLC	w8a16	Snapdragon® 8 Elite Gen 5 Mobile	1.377 ms	2 - 53 MB	NPU
Yolo-v3	TFLITE	float	Snapdragon® 8 Gen 3 Mobile	2.682 ms	0 - 190 MB	NPU
Yolo-v3	TFLITE	float	Qualcomm® QCS8275 (Proxy)	19.961 ms	0 - 146 MB	NPU
Yolo-v3	TFLITE	float	Qualcomm® QCS8550 (Proxy)	3.409 ms	0 - 2 MB	NPU
Yolo-v3	TFLITE	float	Qualcomm® QCS9075	5.661 ms	0 - 30 MB	NPU
Yolo-v3	TFLITE	float	Qualcomm® QCS8450 (Proxy)	9.908 ms	0 - 197 MB	NPU
Yolo-v3	TFLITE	float	Snapdragon® 8 Elite For Galaxy Mobile	2.114 ms	0 - 149 MB	NPU
Yolo-v3	TFLITE	float	Snapdragon® 8 Elite Gen 5 Mobile	1.517 ms	0 - 148 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.