Skimrr VS Open Devdocs

Skimrr's answer

It grew out of a very ordinary problem: the photo folder nobody opens, years of imports nobody sorted, full of duplicates and near-misses nobody's confident enough to delete by hand. (There's a longer writeup on this: https://medium.com/@admin_53263/the-folder-nobody-opens-1b897eb33eed)

Skimrr's answer

Most duplicate-finder tools either ask you to upload photos to a cloud service, rely purely on hashing (missing edited near-duplicates or flagging unrelated similar shots), or use a one-size-fits-all blur cutoff that misfires constantly. Skimrr keeps everything on your machine, adds a local vision-model sanity check most competitors skip entirely, and never deletes anything without a review grid and an undo-able trash.

Skimrr's answer

Every proposed duplicate group is checked by an on-device vision model before it's shown, not just matched by hash, so near-duplicates (cropped, recompressed, lightly edited) get caught without a flood of false positives. The blur score is calibrated per-folder rather than using a fixed threshold, so a soft-focus portrait or a busy, detailed scene doesn't get misjudged. Everything runs 100% locally in Rust, with no account, no cloud, no telemetry, and nothing is ever deleted outright: files go to a reversible local trash first.

Skimrr's answer

Tauri 2 and Rust for the desktop app and the core scanning, hashing, and clustering engine; a local MobileNetV2 ONNX model run through Candle (a pure-Rust ML framework, keeping the whole app free of Python or native ML dependencies) for duplicate verification; and React with TypeScript and i18next for the six-language UI.

Skimrr's answer

People sitting on large personal photo libraries built up over years, with multiple phones, cameras, and import sessions, who want to reclaim disk space without handing their personal photos to a cloud service. Photography enthusiasts, people who regularly import from SD cards, and anyone privacy-conscious about where their photos go.