VectorVFS: Turning Your Filesystem into a Vector Database VectorVFS is a lightweight Python package that transforms your Linux filesystem into a powerful vector database. By utilizing the native Virtual File System (VFS) extended attributes, this tool enables you to store vector embeddings directly alongside each file, turning your existing directory structure into an efficient and semantically searchable data store. One of the standout features of VectorVFS is its support for Meta's Perception Encoders (PE), which includes advanced image and video encoders designed for vision language understanding. These encoders are particularly effective at handling zero-shot image tasks, outperforming other models like InternVL3, Qwen2.5VL, and SigLIP2. Whether you're working with a small collection of files or a vast repository, VectorVFS can handle it, though the initial embedding process may be time-consuming if you're not using a GPU. Key Features In-place Embedding Storage: VectorVFS stores vector embeddings within the filesystem itself, eliminating the need for a separate index or external database. This integration ensures that your embedding store remains organized and aligned with your file structure. Meta’s Perception Encoders: The package supports state-of-the-art encoders from Meta, enhancing its capabilities in vision language understanding and zero-shot image recognition tasks. CPU and GPU Support: VectorVFS works on both CPU and GPU, providing flexibility in how you process and store your data. However, for large collections of images, leveraging a GPU will significantly reduce the initial embedding time. Efficient Querying: Once embedded, files can be searched using semantic queries, making it easier to find relevant content based on context rather than just filename or metadata. Seamless Integration: The tool integrates seamlessly with your existing Linux environment, requiring minimal setup and configuration. This makes it a practical solution for researchers and developers who want to leverage vector embeddings without overhauling their current systems. By harnessing the power of vector embeddings and integrating them directly into your filesystem, VectorVFS offers a unique and efficient solution for managing and searching multimedia data. Its support for advanced encoders and flexible hardware options make it a valuable tool for a wide range of applications, from academic research to industrial projects.