This repository contains a project aimed at removing watermarks from low-resolution sheet music and upscaling it to high-resolution images. The project uses pre-trained deep learning models (UNet and VDSR) to achieve this and includes a graphical user interface (GUI) for scraping, processing, and compiling sheet music into a PDF.
This project aims to remove watermarks from low-resolution sheet music and upscale the images to high resolution. The process involves using a pre-trained UNet model to remove watermarks and a pre-trained VDSR model to enhance the resolution. A GUI is provided to automate the scraping, processing, and compiling of sheet music into a ready-to-use PDF.
The repository includes the following pre-trained models:
These models are provided as state dictionaries and can be found in the models/ directory.
A GUI built with PyQt5 is used to scrape sheet music from a specified website, run it through both the UNet and VDSR models, and compile the processed images into a PDF. This implementation is found in the notebook sheet_music_pyqt5.ipynb.
Clone the repository:
git clone https://github.com/danielshort3/watermark-remover.git
cd watermark-removerInstall the required packages (make sure you have pip and virtualenv installed):
pip install torch torchvision
pip install PyQt5
pip install opencv-python
pip install selenium
pip install webdriver-manager
pip install reportlabEnsure the pre-trained model weights are in the models/ directory.
Launch the GUI by running the sheet_music_pyqt5.ipynb notebook.
Use the GUI to scrape sheet music from a specified website, run it through both the UNet and VDSR models, and compile the processed images into a PDF.