library_name: pytorch
license: other
tags:
- android
pipeline_tag: image-to-image
Real-ESRGAN-x4plus: Optimized for Mobile Deployment
Upscale images and remove image noise
Real-ESRGAN is a machine learning model that upscales an image with minimal loss in quality. The implementation is a derivative of the Real-ESRGAN-x4plus architecture, a larger and more powerful version compared to the Real-ESRGAN-general-x4v3 architecture.
This model is an implementation of Real-ESRGAN-x4plus found here.
This repository provides scripts to run Real-ESRGAN-x4plus on Qualcomm® devices. More details on model performance across various devices, can be found here.
Model Details
- Model Type: Model_use_case.super_resolution
- Model Stats:
- Model checkpoint: RealESRGAN_x4plus
- Input resolution: 128x128
- Number of parameters: 16.7M
- Model size (float): 63.9 MB
- Model size (w8a8): 16.7 MB
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model |
|---|---|---|---|---|---|---|---|---|
| Real-ESRGAN-x4plus | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 449.244 ms | 0 - 319 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 448.83 ms | 0 - 277 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 121.416 ms | 3 - 729 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 117.402 ms | 0 - 694 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 66.756 ms | 3 - 6 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 63.296 ms | 0 - 2 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | ONNX | 70.169 ms | 0 - 44 MB | NPU | Real-ESRGAN-x4plus.onnx.zip |
| Real-ESRGAN-x4plus | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 105.934 ms | 3 - 322 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 105.363 ms | 0 - 277 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 449.244 ms | 0 - 319 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 448.83 ms | 0 - 277 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 61.446 ms | 0 - 3 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 62.895 ms | 0 - 2 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 112.035 ms | 3 - 331 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 111.251 ms | 0 - 296 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 62.984 ms | 0 - 3 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 62.794 ms | 0 - 3 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 105.934 ms | 3 - 322 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 105.363 ms | 0 - 277 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 49.001 ms | 3 - 729 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 48.877 ms | 0 - 685 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 49.819 ms | 6 - 697 MB | NPU | Real-ESRGAN-x4plus.onnx.zip |
| Real-ESRGAN-x4plus | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 38.135 ms | 1 - 300 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 37.926 ms | 0 - 270 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | ONNX | 39.402 ms | 8 - 260 MB | NPU | Real-ESRGAN-x4plus.onnx.zip |
| Real-ESRGAN-x4plus | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | TFLITE | 29.872 ms | 3 - 310 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | QNN_DLC | 24.855 ms | 0 - 276 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | ONNX | 26.593 ms | 7 - 267 MB | NPU | Real-ESRGAN-x4plus.onnx.zip |
| Real-ESRGAN-x4plus | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 64.886 ms | 0 - 0 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 65.611 ms | 38 - 38 MB | NPU | Real-ESRGAN-x4plus.onnx.zip |
| Real-ESRGAN-x4plus | w8a8 | Dragonwing Q-6690 MTP | Qualcomm® Qcm6690 | TFLITE | 376.467 ms | 1 - 422 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | Dragonwing Q-6690 MTP | Qualcomm® Qcm6690 | QNN_DLC | 456.006 ms | 0 - 390 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | TFLITE | 91.91 ms | 1 - 26 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | Dragonwing RB3 Gen 2 Vision Kit | Qualcomm® QCS6490 | QNN_DLC | 109.435 ms | 0 - 3 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 66.186 ms | 1 - 372 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 67.914 ms | 0 - 388 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 38.263 ms | 1 - 604 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 42.507 ms | 0 - 622 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 18.649 ms | 1 - 4 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 25.05 ms | 0 - 3 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 18.468 ms | 1 - 368 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 23.15 ms | 0 - 388 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 1385.344 ms | 0 - 77 MB | GPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 66.186 ms | 1 - 372 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 67.914 ms | 0 - 388 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 18.664 ms | 1 - 5 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 25.197 ms | 0 - 2 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 33.779 ms | 1 - 368 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 37.918 ms | 0 - 394 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 18.7 ms | 1 - 4 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 25.036 ms | 0 - 4 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 18.468 ms | 1 - 368 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 23.15 ms | 0 - 388 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 14.328 ms | 1 - 603 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 16.659 ms | 0 - 613 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | TFLITE | 11.844 ms | 1 - 388 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | Samsung Galaxy S25 | Snapdragon® 8 Elite For Galaxy Mobile | QNN_DLC | 12.581 ms | 0 - 345 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | TFLITE | 35.75 ms | 3 - 390 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | Snapdragon 7 Gen 4 QRD | Snapdragon® 7 Gen 4 Mobile | QNN_DLC | 41.095 ms | 0 - 411 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | TFLITE | 7.465 ms | 1 - 418 MB | NPU | Real-ESRGAN-x4plus.tflite |
| Real-ESRGAN-x4plus | w8a8 | Snapdragon 8 Elite Gen 5 QRD | Snapdragon® 8 Elite Gen5 Mobile | QNN_DLC | 9.112 ms | 0 - 357 MB | NPU | Real-ESRGAN-x4plus.dlc |
| Real-ESRGAN-x4plus | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 26.235 ms | 0 - 0 MB | NPU | Real-ESRGAN-x4plus.dlc |
Installation
Install the package via pip:
# NOTE: 3.10 <= PYTHON_VERSION < 3.14 is supported.
pip install "qai-hub-models[real-esrgan-x4plus]"
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.real_esrgan_x4plus.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.real_esrgan_x4plus.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.real_esrgan_x4plus.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.real_esrgan_x4plus 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.real_esrgan_x4plus.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.real_esrgan_x4plus.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 Real-ESRGAN-x4plus's performance across various devices here. Explore all available models on Qualcomm® AI Hub
License
- The license for the original implementation of Real-ESRGAN-x4plus can be found here.
References
- Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Data
- 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.