LocalLabel – Image & Video Annotation

Label local images and videos for ML: boxes, polygons, masks, tracks, AI assist. Your files never leave your computer.

LocalLabel lets you label images and videos for machine-learning projects without uploading anything. You open a folder on your computer, draw your labels, and save the results as a dataset file. Everything happens inside your browser, so your images and annotations never leave your machine. You can draw boxes (including rotated ones), polygons, lines, points, ellipses and keypoint skeletons, or paint pixel masks with a brush. Each label can have its own attributes, and you can hide, lock or reorder shapes while you work. Undo and redo are unlimited, and keyboard shortcuts make labeling fast. For video, you can step through frames, play the clip, and create tracks that follow an object between keyframes so you don't have to redraw it on every frame. If you have a trained model, you can load it from your disk and let it suggest labels for one image, a whole folder, or a video. The model runs on your own computer. When you're done, you can export your work in the common dataset formats used by popular training tools, and import existing labels to keep working on them. There is no account, no analytics and no network access. Projects save automatically in your browser, and you can also save them into the folder you opened.