An atlas for artificially inflating your datasets

What it does This repository is a heavily bookmarked README acting as a cross-domain literature review. It catalogs data augmentation libraries, GitHub repos, and academic papers—spanning computer vision, NLP, audio, time series, graph data, and even gene expression—complete with a taxonomy diagram that breaks down techniques from affine transforms to edge perturbation. If you need a specialized augmentation method outside the usual image-flipping canon, this is the starting point.

The interesting bit Most lists drown in computer vision; this one stretches into gene expression and graph augmentation without losing structure. It also spotlights the maintainer’s own research on Targeted Data Augmentation, which takes a contrarian swing at bias mitigation by strategically inserting biases rather than scrubbing them out.

Key highlights

• Covers seven distinct areas: computer vision, NLP, audio/speech, time series, graph augmentation, gene expression, and AutoAugment.
• Papers are indexed chronologically from 2015 through 2023 in a separate papers.md.
• Includes practical library descriptions (e.g., albumentations, imgaug, kornia) with feature summaries.
• Provides a taxonomy diagram categorizing techniques from basic jittering to neural style transfer and GANs.
• Highlights the author’s paper on using targeted augmentation to mitigate model bias.

Caveats

• The maintainer is actively looking for help keeping the repository up to date, so some links may grow stale.
• It is a curated index, not a framework; you will find pointers, not importable code.

Verdict Worth a bookmark if you are building a literature review or hunting for domain-specific augmentation ideas beyond standard image transforms. Skip it if you are looking for a single installable package to run augmentations yourself.