- Overview
- Description
- Visual Examples
- Installation
- Usage Instructions
- Citation
- License and Acknowledgement
The PixelPerfect Web App is a sophisticated tool that harnesses the power of AI models to elevate image quality, reduce image artefacts, and enhance image resolution.
Built using cutting-edge technologies, including:
- Django: A high-level Python web framework.
- PyTorch: A leading deep learning framework for AI models.
- Allauth: For secure user authentication and management.
- OAuth2: Ensuring secure authorization.
- Celery: Enabling efficient asynchronous task processing.
- Docker and Docker-compose: Streamlining deployment and ensuring consistency.
Our web app is designed to deliver high-quality results for image restoration tasks. PixelPerfect's commitment to utilizing the latest AI models and advanced technologies guarantees remarkable results in image restoration. Whether you are a professional photographer, designer, or an individual seeking to enhance personal photographs, our application is designed to meet your needs. Elevate your images to the next level of quality and clarity.
Click on the image to view full-size
Login  |
Dashboard  |
Gallery  |
Profile  |
Image Enhancement  |
Result Modal  |
Experience the power of PixelPerfect today and witness the transformation of your images!
Do your low-resolution images lack the detail and clarity you need? PixelPerfect allows you to increase the resolution of your images without compromising quality. Perfect for enhancing old photographs, small thumbnails, or web graphics, this feature breathes new life into your visuals.
Eliminate unwanted image artefacts that can detract from the overall quality of your pictures. Our app employs advanced algorithms to automatically detect and remove noise, compression artefacts, and distortions, ensuring that your images look their best.
After enhancing and upscaling your images, you have the freedom to save the edited versions in two convenient ways:
- Local Storage: Save images directly to your local drive for easy access and offline use.
- Google Drive Integration: Seamlessly sync your edited images with your Google Drive account, making them accessible from anywhere.
Experience enhanced convenience by creating a dedicated user account within the web app. With your account, you can effortlessly save and manage your edited images, streamlining your workflow and facilitating future downloads.
Simplify the registration and login process with our seamless Google OAuth2 integration. You can create an account or log in using your Google credentials, ensuring a secure and frictionless user experience.
Whether you're a professional photographer, graphic designer, or someone looking to improve their personal photos, PixelPerfect provides the tools you need to effortlessly transform your images into sharper, more detailed visuals.
Ready to elevate your images and streamline your image processing workflow? Sign up today or log in with your Google account to get started!
Click on the image to view full-size
 |
 |
 |
 |
* Python 3.9+
* Git
* Docker Desktop
* Docker-compose (Linux)
-
Clone the Repository:
git clone https://github.com/mkrol1997/pixel-perfect
-
Navigate to the Project Directory:
cd pixel-perfect -
Build and Start Containers:
- Make sure you have Docker and Docker Compose installed and running.
docker-compose up --build
-
Access the App:
- Once the containers are running, access the app in your web browser.
- Open your browser and enter the app's URL, which runs at
http://localhost:8000.
-
Shutdown and Cleanup:
docker-compose down
-
Additional Configuration:
- To make sure all of the app features work correctly, it is essential to set all environmental variables listed in the .env file.
Image upscaling increases the resolution of old images by generating additional pixels.
To properly use this functionality, follow these steps:
-
Choose your image input.
-
Select one of the four supported upscaling methods:
-
Resize Scale x2, x3, x4:
- ESPCN
- FSRCNN
- EDSR (Note: This method may take longer compared to others.)
-
Resize Scale x2, x4, x8:
- LapSRN
-
-
Choose the desired scale according to the selected upscaling method.
-
(Optional) You can compress the image quality after upscaling to reduce its size:
- Enable the switch button.
- Choose the compression factor.
-
Save the image using the "Download" button or save it to Google Drive with the "Google Drive" button.
Image enhancement removes all artefacts from your picture. When choosing the enhancement factor, consider the trade-off between artefact removal and preserving fine image details.
To properly remove artefacts, follow these steps:
- Choose your image input.
- Select an image type from the available choices:
- RGB image
- Real-world RGB image
- Greyscale image
- Choose the enhancement factor.
- Save the enhanced image using the "Download" button or save it to Google Drive with the "Google Drive" button.
Full Enhancement combines image upscaling and artefacts removal.
To perform full enhancement, follow these steps:
- Complete the steps for both the "Image Upscaling" and "Image Artefacts Removal" sections mentioned above.
This project includes code from the FBCNN by jiaxi-jiang repository, which is based on the research work:
"Towards Flexible Blind JPEG Artifacts Removal"
- Jiaxi Jiang, Kai Zhang, Radu Timofte Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021
@inproceedings{jiang2021towards,
title={Towards Flexible Blind JPEG Artifacts Removal},
author={Jiang, Jiaxi and Zhang, Kai and Timofte, Radu},
booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
pages={4997--5006},
year={2021}
}
Please visit the original repository for more details on the code and to show your support by giving it a star.
This project is licensed under the MIT License - see the LICENSE file for details.