Image Processing with Deep Learning https://github.com/alizangeneh
This project is a cross-platform desktop application for advanced image processing using deep learning models. The application is built with PyQt5 and works on Windows, Linux, and macOS.
The program includes the following features:
Smart Image Compressor Automatically reduces image resolution to an optimal size while maintaining good visual quality.
Background Removal (AI-based) Uses the rembg deep learning model to remove the background of images.
Face Blurring Privacy Tool Detects faces using a DNN model and applies Gaussian blur to protect privacy.
Quality Booster (Super-Resolution) Enhances image sharpness and resolution by 4x using the RealESRGAN model.
Drag and Drop Support Users can drag and drop images directly into the application window.
Project Folder Structure:
project/
main.py
requirements.txt
README.md
models/
deploy.prototxt
res10_300x300_ssd_iter_140000.caffemodel
RealESRGAN_x4plus.pth
Installation Instructions:
Install all dependencies with: pip install -r requirements.txt
Place the required model files inside the "models" folder:
deploy.prototxt
res10_300x300_ssd_iter_140000.caffemodel
RealESRGAN_x4plus.pth
Running the Application:
python main.py
Optional: Build an Executable File
Windows: pyinstaller --noconsole --onefile main.py
macOS: pyinstaller --onefile main.py
Linux: pyinstaller --onefile main.py
Deep Learning Models Used:
RealESRGAN x4plus: Used for super-resolution.
Rembg (U2Net): Used for background removal.
OpenCV DNN Face Detector (SSD): Used for face detection.
PyTorch CPU: Used for maximum compatibility with older GPUs such as GeForce 210.
Notes:
Torch is installed in CPU mode for full compatibility across systems.
The application is cross-platform.
Due to file size limits, model weights must be downloaded manually.
Author: Ali Zangeneh GitHub: https://github.com/alizangeneh
You must download the following model files from the Release section of this repository:
- RealESRGAN_x4plus.pth
- deploy.prototxt
- res10_300x300_ssd_iter_140000.caffemodel
After downloading, place them inside the models/ folder.