Deep Learning for visually searching and annotating images on a very large scale

Image Search and Annotation on a Very Large Scale
Using Deep Features and MI-File to search and annotate YFCC100M
(Yahoo Flickr Creative Commons 100M) dataset.

VEMIR
Research group

(97 millions images currently indexed)

Choose a random query.

Found 1595445 images similar to your query in 1243 ms (Index access Time: 169 ms, Temporary result managemet time: 687 ms, Database access time: 387 ms)

Query:

Best guessed tags:
Relevance: 9.09, Guessed tags: street
Relevance: 4.13, Guessed tags: car

Search using images similar to your query or see more random images:

(Weight 0) Flickr->	(Weight 0) Flickr->	(Weight 247) Flickr->	(Weight 247) Flickr->	(Weight 280) Flickr->
(Weight 280) Flickr->	(Weight 300) Flickr->	(Weight 300) Flickr->	(Weight 319) Flickr->	(Weight 319) Flickr->
(Weight 330) Flickr->	(Weight 330) Flickr->	(Weight 388) Flickr->	(Weight 388) Flickr->	(Weight 405) Flickr->
(Weight 405) Flickr->	(Weight 465) Flickr->	(Weight 465) Flickr->	(Weight 535) Flickr->	(Weight 535) Flickr->

The above images are miniature versions of the original ones that come from the site www.flickr.com.
All rights are reserved to the author of the original image, which is directly available on the Flickr site through a hypertext link (surface linking).