split-folders

The `split-folders` tool is designed to automate the process of organizing datasets into training, validation, and testing sets, specifically for use in deep learning applications. It supports both class-based and flat directory structures, enabling users to shuffle files, apply randomized oversampling for imbalanced datasets, and split parallel directories while maintaining their alignment. Key features include configurable splitting ratios, reproducible randomness through a seed option, and the ability to group files by prefixes or formats, making it highly adaptable for various dataset management needs.