Hugging Face's logo

qualcomm
/
YOLOv8-Detection

Object Detection
PyTorch
real_time
android
Model card Files Community
1

YOLOv8-Detection: Optimized for Mobile Deployment

Real-time object detection optimized for mobile and edge by Ultralytics

Ultralytics YOLOv8 is a machine learning model that predicts bounding boxes and classes of objects in an image.

This model is an implementation of YOLOv8-Detection found here.

This repository provides scripts to run YOLOv8-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: YOLOv8-N
    • Input resolution: 640x640
    • Number of parameters: 3.18M
    • Model size (float): 12.2 MB
    • Model size (w8a8): 3.25 MB
    • Model size (w8a16): 3.60 MB
Model Precision Device Chipset Target Runtime Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit Target Model
YOLOv8-Detection float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) TFLITE 14.24 ms 0 - 66 MB NPU --
YOLOv8-Detection float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) QNN_DLC 13.359 ms 4 - 96 MB NPU --
YOLOv8-Detection float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) TFLITE 6.584 ms 0 - 40 MB NPU --
YOLOv8-Detection float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) QNN_DLC 8.016 ms 5 - 42 MB NPU --
YOLOv8-Detection float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) TFLITE 4.13 ms 0 - 38 MB NPU --
YOLOv8-Detection float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN_DLC 3.458 ms 0 - 70 MB NPU --
YOLOv8-Detection float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) ONNX 5.481 ms 0 - 100 MB NPU --
YOLOv8-Detection float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) TFLITE 5.589 ms 0 - 65 MB NPU --
YOLOv8-Detection float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) QNN_DLC 5.092 ms 0 - 95 MB NPU --
YOLOv8-Detection float SA7255P ADP Qualcomm® SA7255P TFLITE 14.24 ms 0 - 66 MB NPU --
YOLOv8-Detection float SA7255P ADP Qualcomm® SA7255P QNN_DLC 13.359 ms 4 - 96 MB NPU --
YOLOv8-Detection float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) TFLITE 4.115 ms 0 - 40 MB NPU --
YOLOv8-Detection float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN_DLC 3.464 ms 0 - 81 MB NPU --
YOLOv8-Detection float SA8295P ADP Qualcomm® SA8295P TFLITE 7.643 ms 0 - 34 MB NPU --
YOLOv8-Detection float SA8295P ADP Qualcomm® SA8295P QNN_DLC 7.243 ms 2 - 33 MB NPU --
YOLOv8-Detection float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) TFLITE 4.123 ms 0 - 38 MB NPU --
YOLOv8-Detection float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN_DLC 3.461 ms 0 - 71 MB NPU --
YOLOv8-Detection float SA8775P ADP Qualcomm® SA8775P TFLITE 5.589 ms 0 - 65 MB NPU --
YOLOv8-Detection float SA8775P ADP Qualcomm® SA8775P QNN_DLC 5.092 ms 0 - 95 MB NPU --
YOLOv8-Detection float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile TFLITE 4.13 ms 0 - 39 MB NPU --
YOLOv8-Detection float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN_DLC 3.459 ms 5 - 86 MB NPU --
YOLOv8-Detection float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 5.563 ms 0 - 117 MB NPU --
YOLOv8-Detection float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile TFLITE 3.038 ms 0 - 85 MB NPU --
YOLOv8-Detection float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN_DLC 2.542 ms 5 - 232 MB NPU --
YOLOv8-Detection float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 3.977 ms 0 - 170 MB NPU --
YOLOv8-Detection float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile TFLITE 2.948 ms 0 - 73 MB NPU --
YOLOv8-Detection float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN_DLC 2.371 ms 5 - 135 MB NPU --
YOLOv8-Detection float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 2.964 ms 1 - 89 MB NPU --
YOLOv8-Detection float Snapdragon X Elite CRD Snapdragon® X Elite QNN_DLC 3.898 ms 114 - 114 MB NPU --
YOLOv8-Detection float Snapdragon X Elite CRD Snapdragon® X Elite ONNX 5.453 ms 5 - 5 MB NPU --
YOLOv8-Detection w8a16 QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) QNN_DLC 6.602 ms 1 - 30 MB NPU --
YOLOv8-Detection w8a16 QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) QNN_DLC 3.982 ms 2 - 41 MB NPU --
YOLOv8-Detection w8a16 QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN_DLC 3.276 ms 2 - 14 MB NPU --
YOLOv8-Detection w8a16 QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) ONNX 48.464 ms 14 - 170 MB NPU --
YOLOv8-Detection w8a16 QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) QNN_DLC 3.821 ms 2 - 31 MB NPU --
YOLOv8-Detection w8a16 RB3 Gen 2 (Proxy) Qualcomm® QCS6490 (Proxy) QNN_DLC 13.469 ms 0 - 36 MB NPU --
YOLOv8-Detection w8a16 RB3 Gen 2 (Proxy) Qualcomm® QCS6490 (Proxy) ONNX 172.32 ms 61 - 76 MB CPU --
YOLOv8-Detection w8a16 RB5 (Proxy) Qualcomm® QCS8250 (Proxy) ONNX 151.387 ms 60 - 66 MB CPU --
YOLOv8-Detection w8a16 SA7255P ADP Qualcomm® SA7255P QNN_DLC 6.602 ms 1 - 30 MB NPU --
YOLOv8-Detection w8a16 SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN_DLC 3.278 ms 2 - 13 MB NPU --
YOLOv8-Detection w8a16 SA8295P ADP Qualcomm® SA8295P QNN_DLC 4.47 ms 0 - 32 MB NPU --
YOLOv8-Detection w8a16 SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN_DLC 3.273 ms 2 - 14 MB NPU --
YOLOv8-Detection w8a16 SA8775P ADP Qualcomm® SA8775P QNN_DLC 3.821 ms 2 - 31 MB NPU --
YOLOv8-Detection w8a16 Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN_DLC 3.275 ms 2 - 14 MB NPU --
YOLOv8-Detection w8a16 Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 48.607 ms 14 - 159 MB NPU --
YOLOv8-Detection w8a16 Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN_DLC 2.174 ms 3 - 39 MB NPU --
YOLOv8-Detection w8a16 Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 38.66 ms 2 - 1433 MB NPU --
YOLOv8-Detection w8a16 Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN_DLC 1.87 ms 2 - 45 MB NPU --
YOLOv8-Detection w8a16 Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 41.953 ms 2 - 515 MB NPU --
YOLOv8-Detection w8a16 Snapdragon X Elite CRD Snapdragon® X Elite QNN_DLC 3.636 ms 8 - 8 MB NPU --
YOLOv8-Detection w8a16 Snapdragon X Elite CRD Snapdragon® X Elite ONNX 48.518 ms 24 - 24 MB NPU --
YOLOv8-Detection w8a8 QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) TFLITE 3.333 ms 0 - 23 MB NPU --
YOLOv8-Detection w8a8 QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) QNN_DLC 3.142 ms 1 - 25 MB NPU --
YOLOv8-Detection w8a8 QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) TFLITE 1.662 ms 0 - 37 MB NPU --
YOLOv8-Detection w8a8 QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) QNN_DLC 1.675 ms 1 - 36 MB NPU --
YOLOv8-Detection w8a8 QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) TFLITE 1.493 ms 0 - 15 MB NPU --
YOLOv8-Detection w8a8 QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN_DLC 1.44 ms 1 - 17 MB NPU --
YOLOv8-Detection w8a8 QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) ONNX 5.298 ms 0 - 32 MB NPU --
YOLOv8-Detection w8a8 QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) TFLITE 1.894 ms 0 - 24 MB NPU --
YOLOv8-Detection w8a8 QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) QNN_DLC 1.815 ms 1 - 25 MB NPU --
YOLOv8-Detection w8a8 RB3 Gen 2 (Proxy) Qualcomm® QCS6490 (Proxy) TFLITE 3.805 ms 0 - 31 MB NPU --
YOLOv8-Detection w8a8 RB3 Gen 2 (Proxy) Qualcomm® QCS6490 (Proxy) QNN_DLC 4.815 ms 1 - 32 MB NPU --
YOLOv8-Detection w8a8 RB3 Gen 2 (Proxy) Qualcomm® QCS6490 (Proxy) ONNX 44.018 ms 22 - 38 MB CPU --
YOLOv8-Detection w8a8 RB5 (Proxy) Qualcomm® QCS8250 (Proxy) TFLITE 49.231 ms 2 - 11 MB NPU --
YOLOv8-Detection w8a8 RB5 (Proxy) Qualcomm® QCS8250 (Proxy) ONNX 39.153 ms 16 - 20 MB CPU --
YOLOv8-Detection w8a8 SA7255P ADP Qualcomm® SA7255P TFLITE 3.333 ms 0 - 23 MB NPU --
YOLOv8-Detection w8a8 SA7255P ADP Qualcomm® SA7255P QNN_DLC 3.142 ms 1 - 25 MB NPU --
YOLOv8-Detection w8a8 SA8255 (Proxy) Qualcomm® SA8255P (Proxy) TFLITE 1.5 ms 0 - 15 MB NPU --
YOLOv8-Detection w8a8 SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN_DLC 1.415 ms 1 - 17 MB NPU --
YOLOv8-Detection w8a8 SA8295P ADP Qualcomm® SA8295P TFLITE 2.358 ms 0 - 30 MB NPU --
YOLOv8-Detection w8a8 SA8295P ADP Qualcomm® SA8295P QNN_DLC 2.167 ms 1 - 33 MB NPU --
YOLOv8-Detection w8a8 SA8650 (Proxy) Qualcomm® SA8650P (Proxy) TFLITE 1.496 ms 0 - 15 MB NPU --
YOLOv8-Detection w8a8 SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN_DLC 1.421 ms 1 - 16 MB NPU --
YOLOv8-Detection w8a8 SA8775P ADP Qualcomm® SA8775P TFLITE 1.894 ms 0 - 24 MB NPU --
YOLOv8-Detection w8a8 SA8775P ADP Qualcomm® SA8775P QNN_DLC 1.815 ms 1 - 25 MB NPU --
YOLOv8-Detection w8a8 Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile TFLITE 1.497 ms 0 - 5 MB NPU --
YOLOv8-Detection w8a8 Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN_DLC 1.419 ms 0 - 16 MB NPU --
YOLOv8-Detection w8a8 Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 5.212 ms 0 - 33 MB NPU --
YOLOv8-Detection w8a8 Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile TFLITE 0.99 ms 0 - 36 MB NPU --
YOLOv8-Detection w8a8 Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN_DLC 0.959 ms 1 - 36 MB NPU --
YOLOv8-Detection w8a8 Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 3.781 ms 1 - 89 MB NPU --
YOLOv8-Detection w8a8 Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile TFLITE 0.91 ms 0 - 31 MB NPU --
YOLOv8-Detection w8a8 Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN_DLC 0.844 ms 1 - 37 MB NPU --
YOLOv8-Detection w8a8 Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 3.013 ms 0 - 76 MB NPU --
YOLOv8-Detection w8a8 Snapdragon X Elite CRD Snapdragon® X Elite QNN_DLC 1.651 ms 4 - 4 MB NPU --
YOLOv8-Detection w8a8 Snapdragon X Elite CRD Snapdragon® X Elite ONNX 5.422 ms 3 - 3 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) QNN_DLC 4.455 ms 2 - 27 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN_DLC 2.132 ms 2 - 11 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) ONNX 47.189 ms 27 - 160 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) QNN_DLC 2.661 ms 2 - 27 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 RB3 Gen 2 (Proxy) Qualcomm® QCS6490 (Proxy) ONNX 130.503 ms 56 - 72 MB CPU --
YOLOv8-Detection w8a8_mixed_int16 RB5 (Proxy) Qualcomm® QCS8250 (Proxy) ONNX 123.606 ms 52 - 59 MB CPU --
YOLOv8-Detection w8a8_mixed_int16 SA7255P ADP Qualcomm® SA7255P QNN_DLC 4.455 ms 2 - 27 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN_DLC 2.149 ms 2 - 11 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN_DLC 2.148 ms 2 - 11 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 SA8775P ADP Qualcomm® SA8775P QNN_DLC 2.661 ms 2 - 27 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN_DLC 2.148 ms 2 - 12 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 48.269 ms 13 - 161 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN_DLC 1.445 ms 2 - 37 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 38.324 ms 2 - 1750 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN_DLC 1.248 ms 2 - 31 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 39.884 ms 6 - 521 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 Snapdragon X Elite CRD Snapdragon® X Elite QNN_DLC 2.449 ms 2 - 2 MB NPU --
YOLOv8-Detection w8a8_mixed_int16 Snapdragon X Elite CRD Snapdragon® X Elite ONNX 49.345 ms 25 - 25 MB NPU --

Installation

Install the package via pip:

pip install "qai-hub-models[yolov8-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.yolov8_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.yolov8_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.yolov8_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.yolov8_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.yolov8_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.yolov8_det.demo -- --eval-mode on-device

Deploying compiled model to Android

The models can be deployed using multiple runtimes:

  • TensorFlow Lite (.tflite export): This tutorial provides a guide to deploy the .tflite model in an Android application.

  • QNN (.so export ): This sample app provides instructions on how to use the .so shared library in an Android application.

View on Qualcomm® AI Hub

Get more details on YOLOv8-Detection's performance across various devices here. Explore all available models on Qualcomm® AI Hub

License

  • The license for the original implementation of YOLOv8-Detection can be found here.
  • The license for the compiled assets for on-device deployment can be found here

References

Community

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
Object Detection
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support