README.md

---
library_name: pytorch
license: bsd-3-clause
pipeline_tag: image-classification
tags:
- backbone
- android

---

![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/mnasnet05/web-assets/model_demo.png)

# MNASNet05: Optimized for Mobile Deployment
## Imagenet classifier and general purpose backbone


MNASNet05 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 MNASNet05 found [here](https://github.com/pytorch/vision/blob/main/torchvision/models/mnasnet.py).


This repository provides scripts to run MNASNet05 on Qualcomm® devices.
More details on model performance across various devices, can be found
[here](https://aihub.qualcomm.com/models/mnasnet05).


### Model Details

- **Model Type:** Image classification
- **Model Stats:**
  - Model checkpoint: Imagenet
  - Input resolution: 224x224
  - Number of parameters: 2.21M
  - Model size: 8.45 MB

| Model | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Precision | Primary Compute Unit | Target Model
|---|---|---|---|---|---|---|---|---|
| MNASNet05 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | TFLITE | 0.753 ms | 0 - 1 MB | FP16 | NPU | [MNASNet05.tflite](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.tflite) |
| MNASNet05 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | QNN | 0.817 ms | 1 - 22 MB | FP16 | NPU | [MNASNet05.so](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.so) |
| MNASNet05 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 | ONNX | 0.746 ms | 1 - 145 MB | FP16 | NPU | [MNASNet05.onnx](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.onnx) |
| MNASNet05 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | TFLITE | 0.516 ms | 0 - 49 MB | FP16 | NPU | [MNASNet05.tflite](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.tflite) |
| MNASNet05 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | QNN | 0.56 ms | 0 - 16 MB | FP16 | NPU | [MNASNet05.so](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.so) |
| MNASNet05 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 | ONNX | 0.541 ms | 0 - 53 MB | FP16 | NPU | [MNASNet05.onnx](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.onnx) |
| MNASNet05 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | TFLITE | 0.506 ms | 0 - 22 MB | FP16 | NPU | [MNASNet05.tflite](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.tflite) |
| MNASNet05 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | QNN | 0.552 ms | 0 - 12 MB | FP16 | NPU | Use Export Script |
| MNASNet05 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite | ONNX | 0.467 ms | 0 - 23 MB | FP16 | NPU | [MNASNet05.onnx](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.onnx) |
| MNASNet05 | QCS8550 (Proxy) | QCS8550 Proxy | TFLITE | 0.755 ms | 0 - 1 MB | FP16 | NPU | [MNASNet05.tflite](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.tflite) |
| MNASNet05 | QCS8550 (Proxy) | QCS8550 Proxy | QNN | 0.76 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
| MNASNet05 | SA8255 (Proxy) | SA8255P Proxy | TFLITE | 0.757 ms | 0 - 6 MB | FP16 | NPU | [MNASNet05.tflite](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.tflite) |
| MNASNet05 | SA8255 (Proxy) | SA8255P Proxy | QNN | 0.758 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
| MNASNet05 | SA8775 (Proxy) | SA8775P Proxy | TFLITE | 0.753 ms | 0 - 2 MB | FP16 | NPU | [MNASNet05.tflite](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.tflite) |
| MNASNet05 | SA8775 (Proxy) | SA8775P Proxy | QNN | 0.759 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
| MNASNet05 | SA8650 (Proxy) | SA8650P Proxy | TFLITE | 0.757 ms | 0 - 153 MB | FP16 | NPU | [MNASNet05.tflite](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.tflite) |
| MNASNet05 | SA8650 (Proxy) | SA8650P Proxy | QNN | 0.764 ms | 1 - 2 MB | FP16 | NPU | Use Export Script |
| MNASNet05 | SA8295P ADP | SA8295P | TFLITE | 1.472 ms | 0 - 22 MB | FP16 | NPU | [MNASNet05.tflite](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.tflite) |
| MNASNet05 | SA8295P ADP | SA8295P | QNN | 1.622 ms | 1 - 6 MB | FP16 | NPU | Use Export Script |
| MNASNet05 | QCS8450 (Proxy) | QCS8450 Proxy | TFLITE | 1.024 ms | 0 - 51 MB | FP16 | NPU | [MNASNet05.tflite](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.tflite) |
| MNASNet05 | QCS8450 (Proxy) | QCS8450 Proxy | QNN | 1.107 ms | 1 - 18 MB | FP16 | NPU | Use Export Script |
| MNASNet05 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN | 0.904 ms | 1 - 1 MB | FP16 | NPU | Use Export Script |
| MNASNet05 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 0.841 ms | 7 - 7 MB | FP16 | NPU | [MNASNet05.onnx](https://huggingface.co/qualcomm/MNASNet05/blob/main/MNASNet05.onnx) |




## Installation

This model can be installed as a Python package via pip.

```bash
pip install qai-hub-models
```


## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device

Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) 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.
```bash
qai-hub configure --api_token API_TOKEN
```
Navigate to [docs](https://app.aihub.qualcomm.com/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.

```bash
python -m qai_hub_models.models.mnasnet05.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.mnasnet05.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.

```bash
python -m qai_hub_models.models.mnasnet05.export
```
```
Profiling Results
------------------------------------------------------------
MNASNet05
Device                          : Samsung Galaxy S23 (13)
Runtime                         : TFLITE                 
Estimated inference time (ms)   : 0.8                    
Estimated peak memory usage (MB): [0, 1]                 
Total # Ops                     : 71                     
Compute Unit(s)                 : NPU (71 ops)           
```


## How does this work?

This [export script](https://aihub.qualcomm.com/models/mnasnet05/qai_hub_models/models/MNASNet05/export.py)
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) 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.

```python
import torch

import qai_hub as hub
from qai_hub_models.models.mnasnet05 import 

# Load the model

# Device
device = hub.Device("Samsung Galaxy S23")


```


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.
```python
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.
```python
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](https://myaccount.qualcomm.com/signup).



## Run demo on a cloud-hosted device

You can also run the demo on-device.

```bash
python -m qai_hub_models.models.mnasnet05.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.mnasnet05.demo -- --on-device
```


## Deploying compiled model to Android


The models can be deployed using multiple runtimes:
- TensorFlow Lite (`.tflite` export): [This
  tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
  guide to deploy the .tflite model in an Android application.


- QNN (`.so` export ): This [sample
  app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
provides instructions on how to use the `.so` shared library  in an Android application.


## View on Qualcomm® AI Hub
Get more details on MNASNet05's performance across various devices [here](https://aihub.qualcomm.com/models/mnasnet05).
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)


## License
* The license for the original implementation of MNASNet05 can be found [here](https://github.com/pytorch/vision/blob/main/LICENSE).
* The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)



## References
* [MnasNet: Platform-Aware Neural Architecture Search for Mobile](https://arxiv.org/abs/1807.11626)
* [Source Model Implementation](https://github.com/pytorch/vision/blob/main/torchvision/models/mnasnet.py)



## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).