VAE Image Generator (Dogs Dataset)

A PyTorch-based Variational Autoencoder (VAE) designed to generate and reconstruct images from a custom dog image dataset. This project demonstrates training, generation, and interpolation of images using latent space representations.

---

🚀 Features

• Deep convolutional VAE for RGB images
• Encoder and decoder architecture for 64x64 images
• Custom latent space dimension
• Binary Cross Entropy + KL Divergence loss function
• Image generation and reconstruction
• Latent space interpolation for smooth transitions
• Visualization of loss curves and generated outputs

---

🖼 Example Outputs

• Reconstructed images (per epoch)
• Generated images from latent space
• Latent space interpolation grid
• Loss plots over epochs

---

🛠 Requirements

• Python 3.8+
• PyTorch
• torchvision
• matplotlib
• numpy

Install dependencies:

pip install torch torchvision matplotlib numpy