Awesome-Open-Vocabulary

Why we included this project

Open-vocabulary learning is the problem of getting models to handle classes they never saw during training, and this repo collects the papers and code that make up the field. It grew out of a TPAMI survey, so the organization is thoughtful: detection, segmentation, video, and 3D understanding each get their own section, and every entry points to the paper plus its official implementation. The keyword tags are the most useful part, letting you filter by technique, like caption-based methods or SAM-based ones, which makes it easy to compare approaches that share a recipe. Since the maintainers accept pull requests, the list stays reasonably fresh, and it works well as a map for finding code to study or adapt. Just know it is a reading list, not software you can run.