events and activities such as travel, music, party, festival, football, school; specific locations such as california and italy; scene types such as nature, part, urban, sunset, etc.; the seasons (fall, winter, summer, spring); image style such as portrait, macro, vintage, hdr; and art and culture such as painting, drawing, graffiti, fashion, punk.
Search for any word and then the top retrieval images relevant to that word show up with their scores. This is content-based image search.
This link shows the list of Flickr tags that we trained on, together with our retrieval results, corresponding ImageNet synsets and Flickr tag-based search results.
Here you can find the image classification and retrieval results on NUS-WIDE dataset. This is content-based image classification and retrieval system.
You can upload and image or provide a url and find. Then we run our models on the image and show the top labels with corresponding scores.
Deep Classifiers from Image Tags in the Wild
Hamid Izadinia, Bryan C. Russell, Ali Farhadi, Matthew D. Hoffman, Aaron Hertzmann
Multimedia COMMONS, ACM Multimedia, 2015.
PDF · Demo