This repository contains a Jupyter notebook implementing and evaluating a deep learning-based approach to image super-resolution — the task of enhancing the resolution of low-resolution images.
This was an academic assignment completed as part of a university course on deep learning and computer vision.
- Convolutional Neural Networks (CNNs)
- Bicubic interpolation (baseline)
- Model training on paired low-res / high-res image data
- Evaluation using:
- PSNR (Peak Signal-to-Noise Ratio)
- SSIM (Structural Similarity Index)
The notebook includes visual comparisons between:
- Original high-resolution images
- Low-resolution inputs
- Super-resolved outputs generated by the model
The results demonstrate improved perceptual quality compared to bicubic interpolation.
The notebook uses the following libraries:
- Python 3.x
- TensorFlow / Keras
- NumPy
- Matplotlib
- OpenCV (cv2)
- scikit-image
You can install the dependencies using:
pip install -r requirements.txt