Datasets:

Qdrant

/

google-landmark-geo

like 1

Hi,
It is a great idea to have pre-vectorized image dataset with good license.
Unfortunately, the file with vectors id_payload_vector.json contain only 3266 records (out of 4,132,914 images in original dataset).
For just 3k vectors, one simply do not need a vector database ;)
According to your code, it is possible to make few more requests to wkimedia ecosystem to find much more metadata.