ControlNet: Optimized for Mobile Deployment

Generating visual arts from text prompt and input guiding image

On-device, high-resolution image synthesis from text and image prompts. ControlNet guides Stable-diffusion with provided input image to generate accurate images from given input prompt.

This model is an implementation of ControlNet found here. This repository provides scripts to run ControlNet on Qualcomm® devices. More details on model performance across various devices, can be found here.

Model Details

Model Type: Image generation
Model Stats:
- Input: Text prompt and input image as a reference
- Conditioning Input: Canny-Edge
- QNN-SDK: 2.19
- Text Encoder Number of parameters: 340M
- UNet Number of parameters: 865M
- VAE Decoder Number of parameters: 83M
- ControlNet Number of parameters: 361M
- Model size: 1.4GB

Device	Chipset	Target Runtime	Inference Time (ms)	Peak Memory Range (MB)	Precision	Primary Compute Unit	Target Model
Samsung Galaxy S23 Ultra (Android 13)	Snapdragon® 8 Gen 2	QNN Binary	11.394 ms	0 - 74 MB	UINT16	NPU	TextEncoder_Quantized.bin
Samsung Galaxy S23 Ultra (Android 13)	Snapdragon® 8 Gen 2	QNN Binary	262.52 ms	11 - 17 MB	UINT16	NPU	UNet_Quantized.bin
Samsung Galaxy S23 Ultra (Android 13)	Snapdragon® 8 Gen 2	QNN Binary	390.243 ms	0 - 36 MB	UINT16	NPU	VAEDecoder_Quantized.bin
Samsung Galaxy S23 Ultra (Android 13)	Snapdragon® 8 Gen 2	QNN Binary	100.33 ms	2 - 68 MB	UINT16	NPU	ControlNet_Quantized.bin

Installation

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

pip install "qai-hub-models[controlnet_quantized]"

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 on-device

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.controlnet_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.controlnet_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.controlnet_quantized.export

Profile Job summary of TextEncoder_Quantized
--------------------------------------------------
Device: Samsung Galaxy S24 (14)
Estimated Inference Time: 8.08 ms
Estimated Peak Memory Range: 0.01-137.23 MB
Compute Units: NPU (570) | Total (570)

Profile Job summary of UNet_Quantized
--------------------------------------------------
Device: Samsung Galaxy S24 (14)
Estimated Inference Time: 192.79 ms
Estimated Peak Memory Range: 2.66-1246.59 MB
Compute Units: NPU (5434) | Total (5434)

Profile Job summary of VAEDecoder_Quantized
--------------------------------------------------
Device: Samsung Galaxy S24 (14)
Estimated Inference Time: 294.40 ms
Estimated Peak Memory Range: 0.20-88.33 MB
Compute Units: NPU (409) | Total (409)

Profile Job summary of ControlNet_Quantized
--------------------------------------------------
Device: Samsung Galaxy S24 (14)
Estimated Inference Time: 76.94 ms
Estimated Peak Memory Range: 0.00-532.61 MB
Compute Units: NPU (2406) | Total (2406)

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: Upload compiled model

Upload compiled models from qai_hub_models.models.controlnet_quantized on hub.

import torch

import qai_hub as hub
from qai_hub_models.models.controlnet_quantized import Model

# Load the model
model = Model.from_precompiled()

model_textencoder_quantized = hub.upload_model(model.text_encoder.get_target_model_path())
model_unet_quantized = hub.upload_model(model.unet.get_target_model_path())
model_vaedecoder_quantized = hub.upload_model(model.vae_decoder.get_target_model_path())
model_controlnet_quantized = hub.upload_model(model.controlnet.get_target_model_path())

Step 2: Performance profiling on cloud-hosted device

After uploading compiled 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.


# Device
device = hub.Device("Samsung Galaxy S23")
profile_job_textencoder_quantized = hub.submit_profile_job(
    model=model_textencoder_quantized,
    device=device,
)
profile_job_unet_quantized = hub.submit_profile_job(
    model=model_unet_quantized,
    device=device,
)
profile_job_vaedecoder_quantized = hub.submit_profile_job(
    model=model_vaedecoder_quantized,
    device=device,
)
profile_job_controlnet_quantized = hub.submit_profile_job(
    model=model_controlnet_quantized,
    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_textencoder_quantized = model.text_encoder.sample_inputs()
inference_job_textencoder_quantized = hub.submit_inference_job(
    model=model_textencoder_quantized,
    device=device,
    inputs=input_data_textencoder_quantized,
)
on_device_output_textencoder_quantized = inference_job_textencoder_quantized.download_output_data()

input_data_unet_quantized = model.unet.sample_inputs()
inference_job_unet_quantized = hub.submit_inference_job(
    model=model_unet_quantized,
    device=device,
    inputs=input_data_unet_quantized,
)
on_device_output_unet_quantized = inference_job_unet_quantized.download_output_data()

input_data_vaedecoder_quantized = model.vae_decoder.sample_inputs()
inference_job_vaedecoder_quantized = hub.submit_inference_job(
    model=model_vaedecoder_quantized,
    device=device,
    inputs=input_data_vaedecoder_quantized,
)
on_device_output_vaedecoder_quantized = inference_job_vaedecoder_quantized.download_output_data()

input_data_controlnet_quantized = model.controlnet.sample_inputs()
inference_job_controlnet_quantized = hub.submit_inference_job(
    model=model_controlnet_quantized,
    device=device,
    inputs=input_data_controlnet_quantized,
)
on_device_output_controlnet_quantized = inference_job_controlnet_quantized.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.

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 / .bin export ): This sample app provides instructions on how to use the .so shared library or .bin context binary in an Android application.

View on Qualcomm® AI Hub

Get more details on ControlNet's performance across various devices here. Explore all available models on Qualcomm® AI Hub

License

The license for the original implementation of ControlNet 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.

Usage and Limitations

Model may not be used for or in connection with any of the following applications:

Accessing essential private and public services and benefits;
Administration of justice and democratic processes;
Assessing or recognizing the emotional state of a person;
Biometric and biometrics-based systems, including categorization of persons based on sensitive characteristics;
Education and vocational training;
Employment and workers management;
Exploitation of the vulnerabilities of persons resulting in harmful behavior;
General purpose social scoring;
Law enforcement;
Management and operation of critical infrastructure;
Migration, asylum and border control management;
Predictive policing;
Real-time remote biometric identification in public spaces;
Recommender systems of social media platforms;
Scraping of facial images (from the internet or otherwise); and/or
Subliminal manipulation