wtf is screenshot-search-engine?

Screenshot Search Engine is a fully offline desktop application that indexes your screenshots so you can search through them by their content. It solves a familiar problem: knowing you captured something useful in a screenshot at some point, but having no way to find it again without scrolling through hundreds of images. The app watches a folder you choose and automatically indexes any new screenshots added to it, running the indexing in the background so the interface does not freeze. Each image is read with OCR (Optical Character Recognition, software that pulls text out of a picture) using PyTesseract, with OpenCV used beforehand to clean up the image and improve accuracy on things like app screenshots and UI text. The extracted text, file paths, and a perceptual hash of each image are stored in a local SQLite database, and that hash is used to detect and skip duplicate screenshots. Searching combines exact keyword matches on filenames and extracted text with fuzzy matching, so small typos in a search still find the right result. The interface is a modern, dark mode friendly desktop window built with CustomTkinter, and everything runs locally: no API keys, no cloud services, and no accounts. Because the OCR engine depends on Tesseract-OCR, that has to be installed separately before the app works, and the README's setup instructions are written specifically for Windows. The project can also be packaged into a standalone Windows executable with PyInstaller, though Tesseract itself still needs to be present on the machine running it. This is aimed at someone who takes a lot of screenshots for reference, research, or note-taking and wants to retrieve them by what they actually contain rather than by filename or date. Planned future additions mentioned in the README include semantic search and finding visually similar screenshots, neither of which is built yet.