alex000kim/nsfw_data_scraper

Automates multi-source image scraping across five NSFW/SFW categories using the Ripme tool and public datasets (Danbooru2018, Caltech256), with Docker containerization for reproducibility. Includes data cleaning pipelines that remove corrupted images and automatically stratify collections into train/test splits. Provides a pre-configured fastai CNN training pipeline achieving 91% classification accuracy across the five categories.

12,549 stars. No commits in the last 6 months.