YOLOv10-Detection / README.md

v0.46.1

83508ae verified 7 days ago

7.22 kB

	---
	library_name: pytorch
	license: other
	tags:
	- real_time
	- android
	pipeline_tag: object-detection

	---

	![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolov10_det/web-assets/model_demo.png)

	# YOLOv10-Detection: Optimized for Qualcomm Devices

	Ultralytics YOLOv10 is a machine learning model that predicts bounding boxes and classes of objects in an image.

	This is based on the implementation of YOLOv10-Detection found [here](https://github.com/ultralytics/ultralytics/tree/main/ultralytics/models/yolo/detect).
	This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/yolov10_det) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).

	Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) 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](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/yolov10_det) 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 [YOLOv10-Detection on GitHub](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/yolov10_det) for usage instructions.


	## Model Details

	Model Type: Model_use_case.object_detection

	Model Stats:
	- Model checkpoint: YOLOv10-N
	- Input resolution: 640x640
	- Number of parameters: 2.33M
	- Model size (float): 8.95 MB
	- Model size (w8a8): 2.55 MB
	- Model size (w8a16): 3.04 MB

	## Performance Summary
	\| Model \| Runtime \| Precision \| Chipset \| Inference Time (ms) \| Peak Memory Range (MB) \| Primary Compute Unit
	\|---\|---\|---\|---\|---\|---\|---
	\| YOLOv10-Detection \| ONNX \| float \| Snapdragon® X Elite \| 5.821 ms \| 5 - 5 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| float \| Snapdragon® 8 Gen 3 Mobile \| 3.766 ms \| 5 - 213 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| float \| Qualcomm® QCS8550 (Proxy) \| 5.539 ms \| 5 - 9 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| float \| Qualcomm® QCS9075 \| 7.244 ms \| 5 - 7 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| float \| Snapdragon® 8 Elite For Galaxy Mobile \| 3.472 ms \| 0 - 164 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| float \| Snapdragon® 8 Elite Gen 5 Mobile \| 2.509 ms \| 0 - 156 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Snapdragon® X Elite \| 5.529 ms \| 2 - 2 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Snapdragon® 8 Gen 3 Mobile \| 3.335 ms \| 0 - 165 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Qualcomm® QCS6490 \| 359.505 ms \| 66 - 72 MB \| CPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Qualcomm® QCS8550 (Proxy) \| 5.346 ms \| 1 - 164 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Qualcomm® QCS9075 \| 6.797 ms \| 2 - 5 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Qualcomm® QCM6690 \| 177.917 ms \| 57 - 67 MB \| CPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Snapdragon® 8 Elite For Galaxy Mobile \| 2.557 ms \| 2 - 142 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Snapdragon® 7 Gen 4 Mobile \| 157.609 ms \| 68 - 77 MB \| CPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Snapdragon® 8 Elite Gen 5 Mobile \| 2.294 ms \| 0 - 141 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Snapdragon® X Elite \| 4.417 ms \| 5 - 5 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Snapdragon® 8 Gen 3 Mobile \| 2.939 ms \| 5 - 254 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® QCS8275 (Proxy) \| 13.141 ms \| 3 - 201 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® QCS8550 (Proxy) \| 4.019 ms \| 5 - 12 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® SA8775P \| 5.44 ms \| 1 - 202 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® QCS9075 \| 5.083 ms \| 5 - 11 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® QCS8450 (Proxy) \| 9.216 ms \| 4 - 197 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® SA7255P \| 13.141 ms \| 3 - 201 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® SA8295P \| 8.739 ms \| 1 - 168 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Snapdragon® 8 Elite For Galaxy Mobile \| 2.195 ms \| 5 - 211 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Snapdragon® 8 Elite Gen 5 Mobile \| 1.86 ms \| 5 - 208 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Snapdragon® X Elite \| 4.624 ms \| 2 - 2 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Snapdragon® 8 Gen 3 Mobile \| 2.995 ms \| 0 - 200 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® QCS8275 (Proxy) \| 7.996 ms \| 1 - 174 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® QCS8550 (Proxy) \| 4.248 ms \| 2 - 4 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® SA8775P \| 20.28 ms \| 0 - 173 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® QCS9075 \| 4.734 ms \| 1 - 5 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® QCM6690 \| 20.613 ms \| 2 - 182 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® SA7255P \| 7.996 ms \| 1 - 174 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Snapdragon® 8 Elite For Galaxy Mobile \| 2.075 ms \| 2 - 176 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Snapdragon® 7 Gen 4 Mobile \| 4.406 ms \| 2 - 177 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Snapdragon® 8 Elite Gen 5 Mobile \| 1.767 ms \| 2 - 184 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Snapdragon® 8 Gen 3 Mobile \| 2.647 ms \| 0 - 173 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® QCS8275 (Proxy) \| 12.668 ms \| 0 - 107 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® QCS8550 (Proxy) \| 3.692 ms \| 0 - 3 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® SA8775P \| 5.095 ms \| 0 - 108 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® QCS9075 \| 4.8 ms \| 0 - 13 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® QCS8450 (Proxy) \| 7.759 ms \| 0 - 195 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® SA7255P \| 12.668 ms \| 0 - 107 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® SA8295P \| 8.116 ms \| 0 - 171 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Snapdragon® 8 Elite For Galaxy Mobile \| 1.996 ms \| 0 - 115 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Snapdragon® 8 Elite Gen 5 Mobile \| 1.557 ms \| 0 - 109 MB \| NPU

	## License
	* The license for the original implementation of YOLOv10-Detection can be found
	[here](https://github.com/ultralytics/ultralytics/blob/main/LICENSE).

	## References
	* [YOLOv10: Real-Time End-to-End Object Detection](https://arxiv.org/abs/2405.14458)
	* [Source Model Implementation](https://github.com/ultralytics/ultralytics/tree/main/ultralytics/models/yolo/detect)

	## Community
	* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
	* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).

	---
	library_name: pytorch
	license: other
	tags:
	- real_time
	- android
	pipeline_tag: object-detection

	---

	![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/yolov10_det/web-assets/model_demo.png)

	# YOLOv10-Detection: Optimized for Qualcomm Devices

	Ultralytics YOLOv10 is a machine learning model that predicts bounding boxes and classes of objects in an image.

	This is based on the implementation of YOLOv10-Detection found [here](https://github.com/ultralytics/ultralytics/tree/main/ultralytics/models/yolo/detect).
	This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/yolov10_det) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).

	Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) 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](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/yolov10_det) 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 [YOLOv10-Detection on GitHub](https://github.com/quic/ai-hub-models/blob/main/qai_hub_models/models/yolov10_det) for usage instructions.


	## Model Details

	Model Type: Model_use_case.object_detection

	Model Stats:
	- Model checkpoint: YOLOv10-N
	- Input resolution: 640x640
	- Number of parameters: 2.33M
	- Model size (float): 8.95 MB
	- Model size (w8a8): 2.55 MB
	- Model size (w8a16): 3.04 MB

	## Performance Summary
	\| Model \| Runtime \| Precision \| Chipset \| Inference Time (ms) \| Peak Memory Range (MB) \| Primary Compute Unit
	\|---\|---\|---\|---\|---\|---\|---
	\| YOLOv10-Detection \| ONNX \| float \| Snapdragon® X Elite \| 5.821 ms \| 5 - 5 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| float \| Snapdragon® 8 Gen 3 Mobile \| 3.766 ms \| 5 - 213 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| float \| Qualcomm® QCS8550 (Proxy) \| 5.539 ms \| 5 - 9 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| float \| Qualcomm® QCS9075 \| 7.244 ms \| 5 - 7 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| float \| Snapdragon® 8 Elite For Galaxy Mobile \| 3.472 ms \| 0 - 164 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| float \| Snapdragon® 8 Elite Gen 5 Mobile \| 2.509 ms \| 0 - 156 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Snapdragon® X Elite \| 5.529 ms \| 2 - 2 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Snapdragon® 8 Gen 3 Mobile \| 3.335 ms \| 0 - 165 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Qualcomm® QCS6490 \| 359.505 ms \| 66 - 72 MB \| CPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Qualcomm® QCS8550 (Proxy) \| 5.346 ms \| 1 - 164 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Qualcomm® QCS9075 \| 6.797 ms \| 2 - 5 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Qualcomm® QCM6690 \| 177.917 ms \| 57 - 67 MB \| CPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Snapdragon® 8 Elite For Galaxy Mobile \| 2.557 ms \| 2 - 142 MB \| NPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Snapdragon® 7 Gen 4 Mobile \| 157.609 ms \| 68 - 77 MB \| CPU
	\| YOLOv10-Detection \| ONNX \| w8a16 \| Snapdragon® 8 Elite Gen 5 Mobile \| 2.294 ms \| 0 - 141 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Snapdragon® X Elite \| 4.417 ms \| 5 - 5 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Snapdragon® 8 Gen 3 Mobile \| 2.939 ms \| 5 - 254 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® QCS8275 (Proxy) \| 13.141 ms \| 3 - 201 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® QCS8550 (Proxy) \| 4.019 ms \| 5 - 12 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® SA8775P \| 5.44 ms \| 1 - 202 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® QCS9075 \| 5.083 ms \| 5 - 11 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® QCS8450 (Proxy) \| 9.216 ms \| 4 - 197 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® SA7255P \| 13.141 ms \| 3 - 201 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Qualcomm® SA8295P \| 8.739 ms \| 1 - 168 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Snapdragon® 8 Elite For Galaxy Mobile \| 2.195 ms \| 5 - 211 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| float \| Snapdragon® 8 Elite Gen 5 Mobile \| 1.86 ms \| 5 - 208 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Snapdragon® X Elite \| 4.624 ms \| 2 - 2 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Snapdragon® 8 Gen 3 Mobile \| 2.995 ms \| 0 - 200 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® QCS8275 (Proxy) \| 7.996 ms \| 1 - 174 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® QCS8550 (Proxy) \| 4.248 ms \| 2 - 4 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® SA8775P \| 20.28 ms \| 0 - 173 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® QCS9075 \| 4.734 ms \| 1 - 5 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® QCM6690 \| 20.613 ms \| 2 - 182 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Qualcomm® SA7255P \| 7.996 ms \| 1 - 174 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Snapdragon® 8 Elite For Galaxy Mobile \| 2.075 ms \| 2 - 176 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Snapdragon® 7 Gen 4 Mobile \| 4.406 ms \| 2 - 177 MB \| NPU
	\| YOLOv10-Detection \| QNN_DLC \| w8a16 \| Snapdragon® 8 Elite Gen 5 Mobile \| 1.767 ms \| 2 - 184 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Snapdragon® 8 Gen 3 Mobile \| 2.647 ms \| 0 - 173 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® QCS8275 (Proxy) \| 12.668 ms \| 0 - 107 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® QCS8550 (Proxy) \| 3.692 ms \| 0 - 3 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® SA8775P \| 5.095 ms \| 0 - 108 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® QCS9075 \| 4.8 ms \| 0 - 13 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® QCS8450 (Proxy) \| 7.759 ms \| 0 - 195 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® SA7255P \| 12.668 ms \| 0 - 107 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Qualcomm® SA8295P \| 8.116 ms \| 0 - 171 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Snapdragon® 8 Elite For Galaxy Mobile \| 1.996 ms \| 0 - 115 MB \| NPU
	\| YOLOv10-Detection \| TFLITE \| float \| Snapdragon® 8 Elite Gen 5 Mobile \| 1.557 ms \| 0 - 109 MB \| NPU

	## License
	* The license for the original implementation of YOLOv10-Detection can be found
	[here](https://github.com/ultralytics/ultralytics/blob/main/LICENSE).

	## References
	* [YOLOv10: Real-Time End-to-End Object Detection](https://arxiv.org/abs/2405.14458)
	* [Source Model Implementation](https://github.com/ultralytics/ultralytics/tree/main/ultralytics/models/yolo/detect)

	## Community
	* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
	* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).