library_name: pytorch
license: other
tags:
- real_time
- android
pipeline_tag: object-detection
YOLOv11-Detection: Optimized for Mobile Deployment
Real-time object detection optimized for mobile and edge by Ultralytics
Ultralytics YOLOv11 is a machine learning model that predicts bounding boxes and classes of objects in an image.
This model is an implementation of YOLOv11-Detection found here.
This repository provides scripts to run YOLOv11-Detection on Qualcomm® devices. More details on model performance across various devices, can be found here.
WARNING: The model assets are not readily available for download due to licensing restrictions.
Model Details
- Model Type: Model_use_case.object_detection
- Model Stats:
- Model checkpoint: YOLO11-N
- Input resolution: 640x640
- Number of parameters: 2.64M
- Model size (float): 10.1 MB
- Model size (w8a8): 2.83 MB
- Model size (w8a16): 3.30 MB
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model |
|---|---|---|---|---|---|---|---|---|
| YOLOv11-Detection | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 15.992 ms | 0 - 40 MB | NPU | -- |
| YOLOv11-Detection | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 12.716 ms | 1 - 96 MB | NPU | -- |
| YOLOv11-Detection | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 8.953 ms | 0 - 38 MB | NPU | -- |
| YOLOv11-Detection | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 8.664 ms | 5 - 43 MB | NPU | -- |
| YOLOv11-Detection | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 5.864 ms | 0 - 22 MB | NPU | -- |
| YOLOv11-Detection | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 3.985 ms | 0 - 67 MB | NPU | -- |
| YOLOv11-Detection | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 7.251 ms | 0 - 40 MB | NPU | -- |
| YOLOv11-Detection | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 5.374 ms | 2 - 104 MB | NPU | -- |
| YOLOv11-Detection | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 15.992 ms | 0 - 40 MB | NPU | -- |
| YOLOv11-Detection | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 12.716 ms | 1 - 96 MB | NPU | -- |
| YOLOv11-Detection | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 5.985 ms | 0 - 17 MB | NPU | -- |
| YOLOv11-Detection | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 4.005 ms | 0 - 67 MB | NPU | -- |
| YOLOv11-Detection | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 10.274 ms | 0 - 26 MB | NPU | -- |
| YOLOv11-Detection | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 8.166 ms | 4 - 39 MB | NPU | -- |
| YOLOv11-Detection | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 5.877 ms | 0 - 21 MB | NPU | -- |
| YOLOv11-Detection | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 4.011 ms | 0 - 70 MB | NPU | -- |
| YOLOv11-Detection | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 7.251 ms | 0 - 40 MB | NPU | -- |
| YOLOv11-Detection | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 5.374 ms | 2 - 104 MB | NPU | -- |
| YOLOv11-Detection | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 5.967 ms | 0 - 9 MB | NPU | -- |
| YOLOv11-Detection | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 4.002 ms | 0 - 63 MB | NPU | -- |
| YOLOv11-Detection | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 5.821 ms | 0 - 24 MB | NPU | -- |
| YOLOv11-Detection | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 4.246 ms | 0 - 55 MB | NPU | -- |
| YOLOv11-Detection | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 2.84 ms | 5 - 212 MB | NPU | -- |
| YOLOv11-Detection | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 4.052 ms | 5 - 160 MB | NPU | -- |
| YOLOv11-Detection | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 4.245 ms | 0 - 49 MB | NPU | -- |
| YOLOv11-Detection | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 2.628 ms | 5 - 132 MB | NPU | -- |
| YOLOv11-Detection | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 2.963 ms | 1 - 87 MB | NPU | -- |
| YOLOv11-Detection | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 4.599 ms | 5 - 5 MB | NPU | -- |
| YOLOv11-Detection | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 6.226 ms | 5 - 5 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 7.426 ms | 2 - 33 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 4.544 ms | 2 - 43 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 3.99 ms | 2 - 16 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 4.595 ms | 2 - 35 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 14.698 ms | 2 - 40 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 7.426 ms | 2 - 33 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 3.993 ms | 4 - 15 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 5.266 ms | 1 - 35 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 3.987 ms | 2 - 14 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 4.595 ms | 2 - 35 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 3.976 ms | 2 - 16 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 12.34 ms | 1 - 37 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 2.679 ms | 2 - 41 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 9.461 ms | 2 - 93 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 2.298 ms | 2 - 41 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 8.676 ms | 1 - 97 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 4.695 ms | 0 - 0 MB | NPU | -- |
| YOLOv11-Detection | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 12.586 ms | 2 - 2 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 3.635 ms | 0 - 24 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 3.683 ms | 1 - 28 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 2.022 ms | 0 - 32 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 2.074 ms | 1 - 38 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 1.831 ms | 0 - 12 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 1.787 ms | 1 - 12 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 2.255 ms | 0 - 25 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 2.283 ms | 1 - 29 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 4.068 ms | 0 - 31 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 5.509 ms | 1 - 34 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 60.684 ms | 2 - 13 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 3.635 ms | 0 - 24 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 3.683 ms | 1 - 28 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 1.828 ms | 0 - 11 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 1.806 ms | 1 - 11 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 2.657 ms | 0 - 26 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 2.597 ms | 1 - 30 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 1.828 ms | 0 - 12 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 1.804 ms | 1 - 12 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 2.255 ms | 0 - 25 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 2.283 ms | 1 - 29 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 1.823 ms | 0 - 12 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 1.804 ms | 2 - 13 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 6.738 ms | 0 - 36 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 1.205 ms | 0 - 38 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 1.232 ms | 1 - 39 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 4.987 ms | 1 - 77 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 1.112 ms | 0 - 26 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 1.065 ms | 1 - 34 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 4.279 ms | 1 - 84 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 2.113 ms | 8 - 8 MB | NPU | -- |
| YOLOv11-Detection | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 7.581 ms | 2 - 2 MB | NPU | -- |
Installation
Install the package via pip:
pip install "qai-hub-models[yolov11-det]"
Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
Sign-in to Qualcomm® AI Hub with your
Qualcomm® ID. Once signed in navigate to Account -> Settings -> API Token.
With this API token, you can configure your client to run models on the cloud hosted devices.
qai-hub configure --api_token API_TOKEN
Navigate to docs for more information.
Demo off target
The package contains a simple end-to-end demo that downloads pre-trained weights and runs this model on a sample input.
python -m qai_hub_models.models.yolov11_det.demo
The above demo runs a reference implementation of pre-processing, model inference, and post processing.
NOTE: If you want running in a Jupyter Notebook or Google Colab like environment, please add the following to your cell (instead of the above).
%run -m qai_hub_models.models.yolov11_det.demo
Run model on a cloud-hosted device
In addition to the demo, you can also run the model on a cloud-hosted Qualcomm® device. This script does the following:
- Performance check on-device on a cloud-hosted device
- Downloads compiled assets that can be deployed on-device for Android.
- Accuracy check between PyTorch and on-device outputs.
python -m qai_hub_models.models.yolov11_det.export
How does this work?
This export script leverages Qualcomm® AI Hub to optimize, validate, and deploy this model on-device. Lets go through each step below in detail:
Step 1: Compile model for on-device deployment
To compile a PyTorch model for on-device deployment, we first trace the model
in memory using the jit.trace and then call the submit_compile_job API.
import torch
import qai_hub as hub
from qai_hub_models.models.yolov11_det import Model
# Load the model
torch_model = Model.from_pretrained()
# Device
device = hub.Device("Samsung Galaxy S24")
# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()
pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
# Compile model on a specific device
compile_job = hub.submit_compile_job(
model=pt_model,
device=device,
input_specs=torch_model.get_input_spec(),
)
# Get target model to run on-device
target_model = compile_job.get_target_model()
Step 2: Performance profiling on cloud-hosted device
After compiling models from step 1. Models can be profiled model on-device using the
target_model. Note that this scripts runs the model on a device automatically
provisioned in the cloud. Once the job is submitted, you can navigate to a
provided job URL to view a variety of on-device performance metrics.
profile_job = hub.submit_profile_job(
model=target_model,
device=device,
)
Step 3: Verify on-device accuracy
To verify the accuracy of the model on-device, you can run on-device inference on sample input data on the same cloud hosted device.
input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
model=target_model,
device=device,
inputs=input_data,
)
on_device_output = inference_job.download_output_data()
With the output of the model, you can compute like PSNR, relative errors or spot check the output with expected output.
Note: This on-device profiling and inference requires access to Qualcomm® AI Hub. Sign up for access.
Run demo on a cloud-hosted device
You can also run the demo on-device.
python -m qai_hub_models.models.yolov11_det.demo --eval-mode on-device
NOTE: If you want running in a Jupyter Notebook or Google Colab like environment, please add the following to your cell (instead of the above).
%run -m qai_hub_models.models.yolov11_det.demo -- --eval-mode on-device
Deploying compiled model to Android
The models can be deployed using multiple runtimes:
TensorFlow Lite (
.tfliteexport): This tutorial provides a guide to deploy the .tflite model in an Android application.QNN (
.soexport ): This sample app provides instructions on how to use the.soshared library in an Android application.
View on Qualcomm® AI Hub
Get more details on YOLOv11-Detection's performance across various devices here. Explore all available models on Qualcomm® AI Hub
License
- The license for the original implementation of YOLOv11-Detection can be found here.
- The license for the compiled assets for on-device deployment can be found here
References
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.