Data Filtration

In the Data Filtration section we provide tools for convenient management of your information. Filter, create datasets, and process information with confidence. Your data is always under your control. Using apps, you can set up filtering quickly and easily, ensuring accuracy when working with a variety of information.

Example apps:

• Filter images. App filters images from a project and allows you to copy, move, delete images and assign or remove tags.

• Prompt-based Image Filtering with CLIP. This app allows you to quickly and easily filter and rank images in Supervisely datasets by text prompts. It uses CLIP model to predict the relevance of images to the given text prompt. This app can be useful for filtering or ranking images in a dataset by their content. The relevance (CLIP score) of each image to the given prompt will be shown in a table. The user can choose to filter or sort images by relevance or do both at the same time and then upload images to a new dataset. Tutorial