Hugging Face's logo

qualcomm
/
VIT

Image Classification
PyTorch
LiteRT
ONNX
backbone
android
Model card Files Community
VIT / README.md
qaihm-bot's picture
qaihm-bot
Upload README.md with huggingface_hub
3ea1fa4 verified
preview code
|
raw
history blame contribute delete
11.4 kB
metadata
library_name: pytorch
license: bsd-3-clause
tags:
  - backbone
  - android
pipeline_tag: image-classification

VIT: Optimized for Mobile Deployment

Imagenet classifier and general purpose backbone

VIT 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 VIT found here.

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

Model Details

  • Model Type: Image classification
  • Model Stats:
    • Model checkpoint: Imagenet
    • Input resolution: 224x224
    • Number of parameters: 86.6M
    • Model size: 330 MB
Model Device Chipset Target Runtime Inference Time (ms) Peak Memory Range (MB) Precision Primary Compute Unit Target Model
VIT Samsung Galaxy S23 Snapdragon® 8 Gen 2 TFLITE 13.153 ms 0 - 11 MB FP16 NPU VIT.tflite
VIT Samsung Galaxy S23 Snapdragon® 8 Gen 2 QNN 13.905 ms 1 - 3 MB FP16 NPU VIT.so
VIT Samsung Galaxy S23 Snapdragon® 8 Gen 2 ONNX 14.261 ms 0 - 348 MB FP16 NPU VIT.onnx
VIT Samsung Galaxy S24 Snapdragon® 8 Gen 3 TFLITE 9.051 ms 0 - 283 MB FP16 NPU VIT.tflite
VIT Samsung Galaxy S24 Snapdragon® 8 Gen 3 QNN 9.706 ms 1 - 21 MB FP16 NPU VIT.so
VIT Samsung Galaxy S24 Snapdragon® 8 Gen 3 ONNX 10.037 ms 0 - 73 MB FP16 NPU VIT.onnx
VIT Snapdragon 8 Elite QRD Snapdragon® 8 Elite TFLITE 7.09 ms 0 - 282 MB FP16 NPU VIT.tflite
VIT Snapdragon 8 Elite QRD Snapdragon® 8 Elite QNN 8.915 ms 1 - 292 MB FP16 NPU Use Export Script
VIT Snapdragon 8 Elite QRD Snapdragon® 8 Elite ONNX 9.176 ms 1 - 70 MB FP16 NPU VIT.onnx
VIT SA7255P ADP SA7255P TFLITE 249.895 ms 0 - 279 MB FP16 NPU VIT.tflite
VIT SA7255P ADP SA7255P QNN 249.978 ms 1 - 10 MB FP16 NPU Use Export Script
VIT SA8255 (Proxy) SA8255P Proxy TFLITE 12.769 ms 0 - 12 MB FP16 NPU VIT.tflite
VIT SA8255 (Proxy) SA8255P Proxy QNN 13.918 ms 1 - 3 MB FP16 NPU Use Export Script
VIT SA8295P ADP SA8295P TFLITE 21.062 ms 0 - 279 MB FP16 NPU VIT.tflite
VIT SA8295P ADP SA8295P QNN 20.404 ms 1 - 17 MB FP16 NPU Use Export Script
VIT SA8650 (Proxy) SA8650P Proxy TFLITE 12.91 ms 0 - 30 MB FP16 NPU VIT.tflite
VIT SA8650 (Proxy) SA8650P Proxy QNN 13.938 ms 0 - 2 MB FP16 NPU Use Export Script
VIT SA8775P ADP SA8775P TFLITE 19.796 ms 0 - 279 MB FP16 NPU VIT.tflite
VIT SA8775P ADP SA8775P QNN 20.49 ms 1 - 10 MB FP16 NPU Use Export Script
VIT QCS8275 (Proxy) QCS8275 Proxy TFLITE 249.895 ms 0 - 279 MB FP16 NPU VIT.tflite
VIT QCS8275 (Proxy) QCS8275 Proxy QNN 249.978 ms 1 - 10 MB FP16 NPU Use Export Script
VIT QCS8550 (Proxy) QCS8550 Proxy TFLITE 12.76 ms 0 - 13 MB FP16 NPU VIT.tflite
VIT QCS8550 (Proxy) QCS8550 Proxy QNN 13.817 ms 1 - 4 MB FP16 NPU Use Export Script
VIT QCS9075 (Proxy) QCS9075 Proxy TFLITE 19.796 ms 0 - 279 MB FP16 NPU VIT.tflite
VIT QCS9075 (Proxy) QCS9075 Proxy QNN 20.49 ms 1 - 10 MB FP16 NPU Use Export Script
VIT QCS8450 (Proxy) QCS8450 Proxy TFLITE 18.884 ms 0 - 291 MB FP16 NPU VIT.tflite
VIT QCS8450 (Proxy) QCS8450 Proxy QNN 20.49 ms 1 - 304 MB FP16 NPU Use Export Script
VIT Snapdragon X Elite CRD Snapdragon® X Elite QNN 14.748 ms 1 - 1 MB FP16 NPU Use Export Script
VIT Snapdragon X Elite CRD Snapdragon® X Elite ONNX 14.871 ms 171 - 171 MB FP16 NPU VIT.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.vit.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.vit.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.vit.export

Profiling Results
------------------------------------------------------------
VIT
Device                          : Samsung Galaxy S23 (13)
Runtime                         : TFLITE                 
Estimated inference time (ms)   : 13.2                   
Estimated peak memory usage (MB): [0, 11]                
Total # Ops                     : 1579                   
Compute Unit(s)                 : NPU (1579 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.vit 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.vit.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.vit.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 VIT's performance across various devices here. Explore all available models on Qualcomm® AI Hub

License

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

References

Community