What is it about?
Analysis and retrieval of media collections get more and more challenging the larger the collections become. Traditional relational databases do not support primitives often used in multimedia workloads, such as the nearest-neighbour search on vectors. Cottontail DB is an open-source database management system for multimedia features, designed to help in such situations. It supports traditional relational database operations and text retrieval based on Lucene and, most importantly, efficient vector-space retrieval operations for large datasets.
Featured Image
Photo by Soragrit Wongsa on Unsplash
Why is it important?
Media collections become increasingly huge, which makes it difficult to find relevant content. If content cannot be found, it might as well not exist. Cottontail DB can help tackle this problem by providing an effective and easy to use storage layer for various retrieval applications.
Perspectives
Cottontail DB is significantly faster and more usable than the old database system we were using for our multimedia retrieval system. I think that Cottontail can be a useful tool for researchers working with large amounts of feature data and it is another example of useful and innovative systems research.
Silvan Heller
Read the Original
This page is a summary of: Cottontail DB: An Open Source Database System for Multimedia Retrieval and Analysis, October 2020, ACM (Association for Computing Machinery),
DOI: 10.1145/3394171.3414538.
You can read the full text:
Resources
Contributors
The following have contributed to this page
