Inception-v3-Quantized: Optimized for Mobile Deployment

Quantized Imagenet classifier and general purpose backbone

InceptionNetV3 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 post-training quantized to int8 using samples from Google's open images dataset.

This model is an implementation of Inception-v3-Quantized found here.

This repository provides scripts to run Inception-v3-Quantized 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: 23.9M
- Model size: 23.3 MB

Model	Device	Chipset	Target Runtime	Inference Time (ms)	Peak Memory Range (MB)	Precision	Primary Compute Unit	Target Model
Inception-v3-Quantized	Samsung Galaxy S23	Snapdragon® 8 Gen 2	TFLITE	0.59 ms	0 - 6 MB	INT8	NPU	Inception-v3-Quantized.tflite
Inception-v3-Quantized	Samsung Galaxy S23	Snapdragon® 8 Gen 2	QNN	0.647 ms	0 - 27 MB	INT8	NPU	Inception-v3-Quantized.so
Inception-v3-Quantized	Samsung Galaxy S24	Snapdragon® 8 Gen 3	TFLITE	0.447 ms	0 - 71 MB	INT8	NPU	Inception-v3-Quantized.tflite
Inception-v3-Quantized	Samsung Galaxy S24	Snapdragon® 8 Gen 3	QNN	0.495 ms	0 - 20 MB	INT8	NPU	Inception-v3-Quantized.so
Inception-v3-Quantized	Snapdragon 8 Elite QRD	Snapdragon® 8 Elite	TFLITE	0.418 ms	0 - 24 MB	INT8	NPU	Inception-v3-Quantized.tflite
Inception-v3-Quantized	Snapdragon 8 Elite QRD	Snapdragon® 8 Elite	QNN	0.457 ms	0 - 16 MB	INT8	NPU	Use Export Script
Inception-v3-Quantized	RB3 Gen 2 (Proxy)	QCS6490 Proxy	TFLITE	2.594 ms	0 - 28 MB	INT8	NPU	Inception-v3-Quantized.tflite
Inception-v3-Quantized	RB3 Gen 2 (Proxy)	QCS6490 Proxy	QNN	2.876 ms	0 - 8 MB	INT8	NPU	Use Export Script
Inception-v3-Quantized	RB5 (Proxy)	QCS8250 Proxy	TFLITE	7.709 ms	0 - 2 MB	INT8	NPU	Inception-v3-Quantized.tflite
Inception-v3-Quantized	QCS8550 (Proxy)	QCS8550 Proxy	TFLITE	0.591 ms	0 - 16 MB	INT8	NPU	Inception-v3-Quantized.tflite
Inception-v3-Quantized	QCS8550 (Proxy)	QCS8550 Proxy	QNN	0.651 ms	0 - 1 MB	INT8	NPU	Use Export Script
Inception-v3-Quantized	SA8255 (Proxy)	SA8255P Proxy	TFLITE	0.595 ms	0 - 46 MB	INT8	NPU	Inception-v3-Quantized.tflite
Inception-v3-Quantized	SA8255 (Proxy)	SA8255P Proxy	QNN	0.656 ms	0 - 1 MB	INT8	NPU	Use Export Script
Inception-v3-Quantized	SA8775 (Proxy)	SA8775P Proxy	TFLITE	0.589 ms	0 - 9 MB	INT8	NPU	Inception-v3-Quantized.tflite
Inception-v3-Quantized	SA8775 (Proxy)	SA8775P Proxy	QNN	0.649 ms	0 - 1 MB	INT8	NPU	Use Export Script
Inception-v3-Quantized	SA8650 (Proxy)	SA8650P Proxy	TFLITE	0.591 ms	0 - 247 MB	INT8	NPU	Inception-v3-Quantized.tflite
Inception-v3-Quantized	SA8650 (Proxy)	SA8650P Proxy	QNN	0.654 ms	0 - 2 MB	INT8	NPU	Use Export Script
Inception-v3-Quantized	SA8295P ADP	SA8295P	TFLITE	1.084 ms	0 - 23 MB	INT8	NPU	Inception-v3-Quantized.tflite
Inception-v3-Quantized	SA8295P ADP	SA8295P	QNN	1.294 ms	0 - 6 MB	INT8	NPU	Use Export Script
Inception-v3-Quantized	QCS8450 (Proxy)	QCS8450 Proxy	TFLITE	0.71 ms	0 - 71 MB	INT8	NPU	Inception-v3-Quantized.tflite
Inception-v3-Quantized	QCS8450 (Proxy)	QCS8450 Proxy	QNN	0.765 ms	0 - 22 MB	INT8	NPU	Use Export Script
Inception-v3-Quantized	Snapdragon X Elite CRD	Snapdragon® X Elite	QNN	0.718 ms	0 - 0 MB	INT8	NPU	Use Export Script

Installation

This model can be installed as a Python 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.inception_v3_quantized.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.inception_v3_quantized.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.inception_v3_quantized.export

Profiling Results
------------------------------------------------------------
Inception-v3-Quantized
Device                          : Samsung Galaxy S23 (13)
Runtime                         : TFLITE                 
Estimated inference time (ms)   : 0.6                    
Estimated peak memory usage (MB): [0, 6]                 
Total # Ops                     : 142                    
Compute Unit(s)                 : NPU (142 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.inception_v3_quantized 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.

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.inception_v3_quantized.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.inception_v3_quantized.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 Inception-v3-Quantized's performance across various devices here. Explore all available models on Qualcomm® AI Hub

License

The license for the original implementation of Inception-v3-Quantized can be found here.
The license for the compiled assets for on-device deployment can be found here

References

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.