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: Model_use_case.image_classification
- Model Stats:
- Model checkpoint: Imagenet
- Input resolution: 224x224
- Number of parameters: 86.6M
- Model size (float): 330 MB
- Model size (w8a16): 86.2 MB
- Model size (w8a8): 83.2 MB
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model |
|---|---|---|---|---|---|---|---|---|
| VIT | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 36.611 ms | 0 - 301 MB | NPU | VIT.tflite |
| VIT | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 40.223 ms | 1 - 302 MB | NPU | VIT.dlc |
| VIT | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 12.518 ms | 0 - 312 MB | NPU | VIT.tflite |
| VIT | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 19.663 ms | 0 - 311 MB | NPU | VIT.dlc |
| VIT | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 8.131 ms | 0 - 14 MB | NPU | VIT.tflite |
| VIT | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 11.217 ms | 0 - 25 MB | NPU | VIT.dlc |
| VIT | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 13.392 ms | 1 - 28 MB | NPU | VIT.onnx.zip |
| VIT | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 11.378 ms | 0 - 300 MB | NPU | VIT.tflite |
| VIT | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 14.004 ms | 1 - 302 MB | NPU | VIT.dlc |
| VIT | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 36.611 ms | 0 - 301 MB | NPU | VIT.tflite |
| VIT | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 40.223 ms | 1 - 302 MB | NPU | VIT.dlc |
| VIT | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 8.478 ms | 0 - 21 MB | NPU | VIT.tflite |
| VIT | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 11.205 ms | 3 - 29 MB | NPU | VIT.dlc |
| VIT | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 14.224 ms | 0 - 302 MB | NPU | VIT.tflite |
| VIT | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 17.09 ms | 1 - 307 MB | NPU | VIT.dlc |
| VIT | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 8.464 ms | 0 - 28 MB | NPU | VIT.tflite |
| VIT | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 11.192 ms | 0 - 24 MB | NPU | VIT.dlc |
| VIT | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 11.378 ms | 0 - 300 MB | NPU | VIT.tflite |
| VIT | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 14.004 ms | 1 - 302 MB | NPU | VIT.dlc |
| VIT | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 5.981 ms | 0 - 308 MB | NPU | VIT.tflite |
| VIT | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 7.751 ms | 1 - 310 MB | NPU | VIT.dlc |
| VIT | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 9.091 ms | 0 - 329 MB | NPU | VIT.onnx.zip |
| VIT | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 4.129 ms | 0 - 305 MB | NPU | VIT.tflite |
| VIT | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 5.483 ms | 1 - 308 MB | NPU | VIT.dlc |
| VIT | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 6.362 ms | 1 - 324 MB | NPU | VIT.onnx.zip |
| VIT | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | TFLITE | 3.647 ms | 0 - 302 MB | NPU | VIT.tflite |
| VIT | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | QNN_DLC | 4.23 ms | 1 - 306 MB | NPU | VIT.dlc |
| VIT | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | ONNX | 4.873 ms | 1 - 318 MB | NPU | VIT.onnx.zip |
| VIT | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 11.848 ms | 1127 - 1127 MB | NPU | VIT.dlc |
| VIT | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 13.834 ms | 171 - 171 MB | NPU | VIT.onnx.zip |
| VIT | w8a16 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | ONNX | 1146.481 ms | 45 - 74 MB | CPU | VIT.onnx.zip |
| VIT | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 300.518 ms | 36 - 129 MB | NPU | VIT.onnx.zip |
| VIT | w8a16 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 534.357 ms | 86 - 100 MB | CPU | VIT.onnx.zip |
| VIT | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 228.39 ms | 60 - 96 MB | NPU | VIT.onnx.zip |
| VIT | w8a16 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 342.586 ms | 59 - 99 MB | NPU | VIT.onnx.zip |
| VIT | w8a16 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | ONNX | 592.711 ms | 91 - 111 MB | CPU | VIT.onnx.zip |
| VIT | w8a16 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | ONNX | 186.6 ms | 60 - 103 MB | NPU | VIT.onnx.zip |
| VIT | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 173.175 ms | 71 - 71 MB | NPU | VIT.onnx.zip |
| VIT | w8a8 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | TFLITE | 71.68 ms | 1 - 99 MB | NPU | VIT.tflite |
| VIT | w8a8 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | ONNX | 867.452 ms | 66 - 99 MB | CPU | VIT.onnx.zip |
| VIT | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 15.57 ms | 0 - 48 MB | NPU | VIT.tflite |
| VIT | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 7.99 ms | 0 - 56 MB | NPU | VIT.tflite |
| VIT | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 7.435 ms | 0 - 78 MB | NPU | VIT.tflite |
| VIT | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 269.228 ms | 25 - 115 MB | NPU | VIT.onnx.zip |
| VIT | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 7.776 ms | 0 - 47 MB | NPU | VIT.tflite |
| VIT | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 467.975 ms | 62 - 78 MB | CPU | VIT.onnx.zip |
| VIT | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 15.57 ms | 0 - 48 MB | NPU | VIT.tflite |
| VIT | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 7.442 ms | 0 - 33 MB | NPU | VIT.tflite |
| VIT | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 9.722 ms | 0 - 49 MB | NPU | VIT.tflite |
| VIT | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 7.475 ms | 0 - 19 MB | NPU | VIT.tflite |
| VIT | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 7.776 ms | 0 - 47 MB | NPU | VIT.tflite |
| VIT | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 5.24 ms | 0 - 53 MB | NPU | VIT.tflite |
| VIT | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 233.46 ms | 60 - 98 MB | NPU | VIT.onnx.zip |
| VIT | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 4.067 ms | 0 - 53 MB | NPU | VIT.tflite |
| VIT | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 226.962 ms | 60 - 95 MB | NPU | VIT.onnx.zip |
| VIT | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | TFLITE | 20.662 ms | 1 - 32 MB | NPU | VIT.tflite |
| VIT | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | ONNX | 463.437 ms | 61 - 84 MB | CPU | VIT.onnx.zip |
| VIT | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | TFLITE | 3.462 ms | 0 - 56 MB | NPU | VIT.tflite |
| VIT | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | ONNX | 181.858 ms | 60 - 95 MB | NPU | VIT.onnx.zip |
| VIT | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 151.1 ms | 69 - 69 MB | NPU | VIT.onnx.zip |
| VIT | w8a8_mixed_int16 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | ONNX | 892.433 ms | 83 - 124 MB | CPU | VIT.onnx.zip |
| VIT | w8a8_mixed_int16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 342.28 ms | 33 - 139 MB | NPU | VIT.onnx.zip |
| VIT | w8a8_mixed_int16 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | ONNX | 457.866 ms | 85 - 106 MB | CPU | VIT.onnx.zip |
| VIT | w8a8_mixed_int16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 288.041 ms | 79 - 124 MB | NPU | VIT.onnx.zip |
| VIT | w8a8_mixed_int16 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 250.701 ms | 79 - 121 MB | NPU | VIT.onnx.zip |
| VIT | w8a8_mixed_int16 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | ONNX | 489.289 ms | 85 - 109 MB | CPU | VIT.onnx.zip |
| VIT | w8a8_mixed_int16 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | ONNX | 238.335 ms | 77 - 120 MB | NPU | VIT.onnx.zip |
| VIT | w8a8_mixed_int16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 237.212 ms | 137 - 137 MB | NPU | VIT.onnx.zip |
Installation
Install the package via pip:
pip install qai-hub-models
Configure Qualcomm® AI Hub Workbench to run this model on a cloud-hosted device
Sign-in to Qualcomm® AI Hub Workbench 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
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 S25")
# 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 Workbench. 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 --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.vit.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 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
- An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
- Source Model Implementation
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.
- Downloads last month
- 353