synderm documentation

Synderm is a package designed to enhance image classification tasks using synthetic data generation. It provides tools to generate high-quality synthetic images using diffusion models, fine-tune these models on your specific datasets, and seamlessly integrate synthetic data into your training pipelines to improve classifier performance.

Contents:

• Vignette: Using the Synderm dataset
  • Dataset Statistics
  • Dataset Details
• Vignette: Generating Synthetic Images for Training Your Classifier
  • Introduction
  • 1. Creating the dataset
  • 2. Train the synthetic image generator
  • 3. Generate synthetic images
  • 4. Training the classifier
  • Conclusion
• Vignette: Augmenting Your Classifier with Synthetic Images
  • Introduction
  • 1. Load your dataset
  • 2. Loading the synthetic images from HuggingFace
  • 3. Mixing the real and synthetic training images
  • 4. Model training and validation
• Modules
  • fine_tune package
  • generation package
  • utils package

Indices and tables

• Index

• Module Index

• Search Page