Name
Abstract | ||
|---|---|---|
This paper proposes direct learning of image classification from image tags in the wild, without filtering. Each wild tag is supplied by the user who shared the image online. Enormous numbers of these tags are freely available, and they give insight about the image categories important to users and to image classification. Our main contribution is an analysis of the Flickr 100 Million Image dataset, including several useful observations about the statistics of these tags. We introduce a large-scale robust classification algorithm, in order to handle the inherent noise in these tags, and a calibration procedure to better predict objective annotations. We show that freely available, wild tag can obtain similar or superior results to large databases of costly manual annotations. |
Year | DOI | Venue |
|---|---|---|
2015 | 10.1145/2814815.2814821 | MMCommons@ACM Multimedia |
Keywords | Field | DocType |
Deep Learning,Tags in the Wild,Large-scale Robust Classification,Image Tag Suggestion,Image Retrieval | Data mining,Information retrieval,Computer science,Image retrieval,Filter (signal processing),Artificial intelligence,Deep learning,Contextual image classification | Conference |
Citations | PageRank | References |
15 | 0.56 | 20 |
Authors | ||
5 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Hamid Izadinia | 1 | 164 | 11.16 |
| Bryan C. Russell | 2 | 2570 | 217.78 |
| Ali Farhadi | 3 | 4492 | 190.40 |
| Matthew D. Hoffman | 4 | 1117 | 60.21 |
| Aaron Hertzmann | 5 | 6002 | 352.67 |