Hugging Face's logo

qualcomm
/
Swin-Base

Image Classification
PyTorch
LiteRT
ONNX
backbone
android
Model card Files Community
Swin-Base / README.md
qaihm-bot's picture
qaihm-bot
Upload README.md with huggingface_hub
55284e4 verified
preview code
|
raw
history blame contribute delete
13.1 kB
metadata
library_name: pytorch
license: other
tags:
  - backbone
  - android
pipeline_tag: image-classification

Swin-Base: Optimized for Mobile Deployment

Imagenet classifier and general purpose backbone

SwinBase is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.

This model is an implementation of Swin-Base found here.

This repository provides scripts to run Swin-Base on Qualcomm® devices. More details on model performance across various devices, can be found here.

Model Details

  • Model Type: Model_use_case.image_classification
  • Model Stats:
    • Model checkpoint: Imagenet
    • Input resolution: 224x224
    • Number of parameters: 88.8M
    • Model size (float): 339 MB
Model Precision Device Chipset Target Runtime Inference Time (ms) Peak Memory Range (MB) Primary Compute Unit Target Model
Swin-Base float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) TFLITE 62.202 ms 0 - 372 MB NPU Swin-Base.tflite
Swin-Base float QCS8275 (Proxy) Qualcomm® QCS8275 (Proxy) QNN 59.662 ms 0 - 9 MB NPU Use Export Script
Swin-Base float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) TFLITE 29.353 ms 0 - 379 MB NPU Swin-Base.tflite
Swin-Base float QCS8450 (Proxy) Qualcomm® QCS8450 (Proxy) QNN 39.401 ms 1 - 460 MB NPU Use Export Script
Swin-Base float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) TFLITE 22.379 ms 0 - 34 MB NPU Swin-Base.tflite
Swin-Base float QCS8550 (Proxy) Qualcomm® QCS8550 (Proxy) QNN 21.551 ms 1 - 3 MB NPU Use Export Script
Swin-Base float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) TFLITE 25.826 ms 0 - 372 MB NPU Swin-Base.tflite
Swin-Base float QCS9075 (Proxy) Qualcomm® QCS9075 (Proxy) QNN 24.76 ms 1 - 10 MB NPU Use Export Script
Swin-Base float SA7255P ADP Qualcomm® SA7255P TFLITE 62.202 ms 0 - 372 MB NPU Swin-Base.tflite
Swin-Base float SA7255P ADP Qualcomm® SA7255P QNN 59.662 ms 0 - 9 MB NPU Use Export Script
Swin-Base float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) TFLITE 22.254 ms 0 - 26 MB NPU Swin-Base.tflite
Swin-Base float SA8255 (Proxy) Qualcomm® SA8255P (Proxy) QNN 21.644 ms 1 - 3 MB NPU Use Export Script
Swin-Base float SA8295P ADP Qualcomm® SA8295P TFLITE 32.21 ms 0 - 361 MB NPU Swin-Base.tflite
Swin-Base float SA8295P ADP Qualcomm® SA8295P QNN 30.655 ms 1 - 18 MB NPU Use Export Script
Swin-Base float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) TFLITE 22.199 ms 0 - 24 MB NPU Swin-Base.tflite
Swin-Base float SA8650 (Proxy) Qualcomm® SA8650P (Proxy) QNN 21.806 ms 1 - 3 MB NPU Use Export Script
Swin-Base float SA8775P ADP Qualcomm® SA8775P TFLITE 25.826 ms 0 - 372 MB NPU Swin-Base.tflite
Swin-Base float SA8775P ADP Qualcomm® SA8775P QNN 24.76 ms 1 - 10 MB NPU Use Export Script
Swin-Base float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile TFLITE 22.273 ms 0 - 23 MB NPU Swin-Base.tflite
Swin-Base float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile QNN 21.787 ms 0 - 31 MB NPU Use Export Script
Swin-Base float Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 19.71 ms 0 - 382 MB NPU Swin-Base.onnx
Swin-Base float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile TFLITE 15.874 ms 0 - 378 MB NPU Swin-Base.tflite
Swin-Base float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile QNN 15.104 ms 1 - 378 MB NPU Use Export Script
Swin-Base float Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 13.584 ms 1 - 377 MB NPU Swin-Base.onnx
Swin-Base float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile TFLITE 14.567 ms 0 - 372 MB NPU Swin-Base.tflite
Swin-Base float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile QNN 13.41 ms 1 - 329 MB NPU Use Export Script
Swin-Base float Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 12.472 ms 1 - 332 MB NPU Swin-Base.onnx
Swin-Base float Snapdragon X Elite CRD Snapdragon® X Elite QNN 22.081 ms 1 - 1 MB NPU Use Export Script
Swin-Base float Snapdragon X Elite CRD Snapdragon® X Elite ONNX 23.173 ms 175 - 175 MB NPU Swin-Base.onnx
Swin-Base w8a16 Samsung Galaxy S23 Snapdragon® 8 Gen 2 Mobile ONNX 175.884 ms 518 - 730 MB NPU Swin-Base.onnx
Swin-Base w8a16 Samsung Galaxy S24 Snapdragon® 8 Gen 3 Mobile ONNX 134.517 ms 662 - 960 MB NPU Swin-Base.onnx
Swin-Base w8a16 Snapdragon 8 Elite QRD Snapdragon® 8 Elite Mobile ONNX 127.068 ms 649 - 914 MB NPU Swin-Base.onnx
Swin-Base w8a16 Snapdragon X Elite CRD Snapdragon® X Elite ONNX 172.221 ms 918 - 918 MB NPU Swin-Base.onnx

Installation

Install the package via pip:

pip install qai-hub-models

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.swin_base.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.swin_base.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.swin_base.export

Profiling Results
------------------------------------------------------------
Swin-Base
Device                          : cs_8275 (ANDROID 14)                  
Runtime                         : TFLITE                                
Estimated inference time (ms)   : 62.2                                  
Estimated peak memory usage (MB): [0, 372]                              
Total # Ops                     : 1568                                  
Compute Unit(s)                 : npu (1568 ops) gpu (0 ops) cpu (0 ops)

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.swin_base 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.swin_base.demo --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.swin_base.demo -- --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 Swin-Base's performance across various devices here. Explore all available models on Qualcomm® AI Hub

License

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

References

Community