Enhancing Images with AI: My New App Experiment

A Simple Idea Born from the MIT AI Course

As part of the MIT online course “Designing and Developing AI Products and Services”, I created a simple application that uses artificial intelligence to enhance and restore images. The goal was to explore how AI models can improve visual quality by recovering details and sharpening low-resolution photos.

What started as a simple exercise to put into practice my knowledge and stretch my design skills became an amazing project that combines creativity, coding, and machine learning.

How it Works

The app is built in Python and uses Real-ESRGAN, an open-source model specialized in image super-resolution essentially, making small or blurry images look sharp again.

Here’s how it works step by step:

1. Initialize the Real-ESRGAN model using Torch, the deep learning framework that powers many of today’s AI applications. Torch loads the pre-trained weights and prepares the model to process images efficiently on CPU or GPU.
2. Process the image through the Real-ESRGAN model, which enhances edges, textures, and fine details.
3. Display and save the improved version using Matplotlib to visualize the before-and-after effect.

The results are remarkable. Even heavily compressed or pixelated photos regain clarity and structure, revealing how far AI-based enhancement has come.

The Learning Journey

Building this project helped me understand:

• How AI models like ESRGAN reconstruct lost information in images
• The practical use of pre-trained networks for visual tasks
• How to manage and visualize image data in Python

Beyond the technical lessons, I enjoyed the creative process — experimenting, testing, and seeing tangible improvements on screen. It reminded me that learning AI isn’t just about code, but about finding real-world problems and crafting solutions that make a difference.

What’s Next

My next step is to develop a simple web interface that allows anyone to upload an image and experience the enhancement process directly. This could turn the app into a useful demonstration of applied AI in everyday contexts.

Working on this project reinforced an important idea: the best way to learn AI is by building. Every small experiment adds up, and this one reminded me how rewarding it is to turn code into something visual, practical, and creative.