aperturedata/aperturedb-standalone repository overview

ApertureDB is a specialized database designed to simplify data access for computer vision, machine learning, and vector search use cases. You can read the details of its architecture and query specification on our documentation⁠. Blogs describing various use cases, our vision, and design choices can be found on our website⁠.

⁠ApertureDB Core Components

• Data Management:
  • Supports native management of images, videos, documents, or any blob types
  • Provides visual preprocessing operations like zoom, crop, sampling, and thumbnail creation as part of the query API
• Vector Database:
  • Enables similarity search over n-dimensional feature vectors or embeddings
  • Built on top of Facebook’s FAISS library with durability and scalability
• Graph-Based Metadata Filtering:
  • Uses an in-memory graph database to store application metadata as a knowledge graph
  • Supports bounding boxes and regions of interest for labeling or annotation-based searches.
  • Represents images, videos, blobs, annotations, and embeddings within this metadata.
• Unified API:
  • Redirects user queries to the right components internally.
  • Exposes a unified JSON-based native query language to ML pipelines and end users.
• Backing Store:
  • Stores and accesses data from disks mapped to the ApertureDB server or cloud object stores.
  • Allows queries to add, modify, and search multimodal data, metadata, annotations, and feature vectors.

⁠Getting Started

⁠Aperture-hosted Trials

You can do a quick trial of ApertureDB using one of our demo examples⁠.

We currently offer a choice of pre-loaded datasets which lend themselves to specific use-cases. These use-cases are also provided as a collection of notebooks in jupyterlab environment.

This requires no setup, and is the quickest way to get interacting with ApertureDB!

⁠Download and Setup This Docker With or Without UI

We have packaged ApertureDB and all the dependencies as well as ApertureDB frontend / UI, in a set of docker images. Please follow the documentation and examples⁠. Right now, the docker images work on Linux x86-64 machines and install a single-machine instance of ApertureDB. For a distributed deployment, reach out to us at [email protected]⁠.

You can connect to the ApertureDB server using our Python client available via pip install⁠. If you choose to install the UI using docker compose, you can connect to the port 80 of the machine you are running the database on. The default database password is "admin:admin". We strongly recommend changing the password on first use

⁠Contacts

• Have questions? Join our slack channel⁠ or email [email protected]⁠
• Want to stay in touch with latest releases? Subscribe to our Newsletters⁠ or follow ApertureData on LinkedIn⁠
• Looking for a managed cloud? Check pricing, need something personalised? We're at [email protected]⁠

⁠License

View license information⁠ for the software contained in this image. If you have any questions about use of this image, contact ApertureData at [email protected]⁠.

As with all Docker images, these likely also contain other software which may be under other licenses (such as FFMPEG library, etc from the base distribution, along with any direct or indirect dependencies of the primary software being contained).

As for any pre-built image usage, it is the image user's responsibility to ensure that any use of this image complies with any relevant licenses for all software contained within.