This repository houses the code used to create an Annoy Index of CLIP embedded images. The CLIP model was used via the Sentence Transformer library. The images used in this repository are medieval manuscripts and were gathered from E-Codices.
This repository also has the code for querying the Annoy index with an encoded text-based query. This allows a user to search a database of images with text.
Video link: https://youtu.be/x8C8IMqHxFA
For a complete citation to all images used in this repository, see data/citation.md.