Generative AI projects

This repository contains five Generative Artificial Intelligence (AI) tasks implemented using Python and Deep Learning.

The project showcases my practical understanding of Generative AI techniques such as text generation, image generation, probabilistic models, GANs, and neural style transfer.

📌 Tasks Overview (With Details)

🔹 Task 01: Text Generation using GPT-2

Description:
This task uses a pre-trained GPT-2 transformer model to generate human-like text. A short input sentence (prompt) is given, and the model continues the text in a meaningful way.

Concepts Used:

• Natural Language Processing (NLP)
• Transformers
• Text generation

---

🔹 Task 02: Image Generation from Text

Description:
In this task, images are generated from text prompts using diffusion-based generative models.
The model converts a descriptive sentence into a visual image.

Concepts Used:

• Diffusion models
• Text-to-image generation
• Creative AI

---

🔹 Task 03: Text Generation using Markov Chains

Description:
This task demonstrates a simple probability-based text generation method. Each word is generated based on the previous word using a Markov process.

Concepts Used:

• Markov Chains
• Probabilistic modeling
• Basic text generation

---

🔹 Task 04: Image-to-Image Translation using pix2pix (cGAN)

Description:
This task uses a Conditional GAN (pix2pix) to translate one image into another. For example:

• Sketch → Realistic image
• Map → Satellite image

Concepts Used:

• Generative Adversarial Networks (GANs)
• Conditional GANs
• Image translation

---

🔹 Task 05: Neural Style Transfer

Description:
This task applies the artistic style of one image to the content of another image using a pre-trained VGG19 convolutional neural network.

Example:

• Content image: Photograph
• Style image: Painting

Concepts Used:

• CNNs
• Feature extraction
• Artistic image generation

🛠 Requirements (Libraries Used)

torch torchvision transformers diffusers numpy matplotlib pillow tqdm scipy

🛠 Tools Used

The following tools and technologies were used to build and run this project:

🔹 Programming Language

• Python (3.8+) – Used for writing all scripts and implementations

🔹 Development Tools

• Google Colab – Cloud-based execution with GPU support

▶ Where to Run This Project

Run on Google Colab (Recommended) Best option if your system is slow or you need GPU support.

Why Colab?

• Free GPU access
• No installation issues
• Easy to run deep learning code

Steps:

1. Open https://colab.research.google.com
2. Upload the project files or individual task scripts
3. Enable GPU:
   • Runtime → Change runtime type → Select GPU
4. Run the code cells

---

✅ Recommended Choice

• Google Colab for heavy tasks like Image Generation and Neural Style Transfer
• Local system for lighter tasks like Markov Chain and GPT-2 text generation

👤 Author

Deepak Methre