ImageSorter

ImageSorter is a file organizer tool. A free, no-sign-up web tool that uses OpenAI's CLIP zero-shot model to classify and drag-and-drop sort batches of images into custom tags, with an adjustable confidence threshold and tag-order sorting.

ImageSorter is a browser-based tool for bringing order to a folder of unsorted images. The premise is deliberately plain: you load a batch of pictures, define the tags you care about, and the tool moves each image into the right bucket. Sorting happens through a drag-and-drop interface, so the interaction model will feel familiar to anyone who has ever tidied files by hand. The difference is that ImageSorter can pre-sort on your behalf rather than leaving every decision to you. The engine behind that automation is OpenAI's CLIP zero-shot image classification model. Zero-shot means you are not restricted to a fixed set of labels baked in by the developer. You supply the tag names yourself (the homepage demonstrates Dog, House, Insect and Fish), and the model scores each image against those labels. A confidence threshold slider lets you decide how sure the model must be before it assigns a picture to a category, and a Predict control triggers the classification pass. A sort-by-tag-order option lets you control how the resulting groups are arranged. The audience is anyone sitting on an unlabeled photo pile: photographers culling shoots, families organizing years of camera roll exports, researchers assembling small visual datasets, or teams that need a quick first pass before human review. Because no sign up is required, the barrier to trying it is essentially zero. Where ImageSorter fits is the lightweight end of the image management market. It is not a digital asset manager with metadata schemas, versioning or team permissions, and it is not a photo editor. It is a focused sorting utility that leans on a general-purpose vision model instead of custom-trained classifiers. For small to medium batches where you want categories rather than a full library system, that focus is the point.