%0 Conference Paper %1 plummer2015flickr30k %A Plummer, B. A. %A Wang, L. %A Cervantes, C. M. %A Caicedo, J. C. %A Hockenmaier, J. %A Lazebnik, S. %B 2015 IEEE International Conference on Computer Vision (ICCV) %D 2015 %K flickr localisation nlp object paper-dzhi text %P 2641-2649 %R 10.1109/ICCV.2015.303 %T Flickr30k Entities: Collecting Region-to-Phrase Correspondences for Richer Image-to-Sentence Models %U https://ieeexplore.ieee.org/document/7410660 %X The Flickr30k dataset has become a standard benchmark for sentence-based image description. This paper presents Flickr30k Entities, which augments the 158k captions from Flickr30k with 244k coreference chains linking mentions of the same entities in images, as well as 276k manually annotated bounding boxes corresponding to each entity. Such annotation is essential for continued progress in automatic image description and grounded language understanding. We present experiments demonstrating the usefulness of our annotations for text-to-image reference resolution, or the task of localizing textual entity mentions in an image, and for bidirectional image-sentence retrieval. These experiments confirm that we can further improve the accuracy of state-of-the-art retrieval methods by training with explicit region-to-phrase correspondence, but at the same time, they show that accurately inferring this correspondence given an image and a caption remains really challenging.

@inproceedings{plummer2015flickr30k, abstract = {The Flickr30k dataset has become a standard benchmark for sentence-based image description. This paper presents Flickr30k Entities, which augments the 158k captions from Flickr30k with 244k coreference chains linking mentions of the same entities in images, as well as 276k manually annotated bounding boxes corresponding to each entity. Such annotation is essential for continued progress in automatic image description and grounded language understanding. We present experiments demonstrating the usefulness of our annotations for text-to-image reference resolution, or the task of localizing textual entity mentions in an image, and for bidirectional image-sentence retrieval. These experiments confirm that we can further improve the accuracy of state-of-the-art retrieval methods by training with explicit region-to-phrase correspondence, but at the same time, they show that accurately inferring this correspondence given an image and a caption remains really challenging.}, added-at = {2020-01-20T10:04:10.000+0100}, author = {{Plummer}, B. A. and {Wang}, L. and {Cervantes}, C. M. and {Caicedo}, J. C. and {Hockenmaier}, J. and {Lazebnik}, S.}, biburl = {https://www.bibsonomy.org/bibtex/2d7874dca41004be97d7fc817fa9477c7/nosebrain}, booktitle = {2015 IEEE International Conference on Computer Vision (ICCV)}, description = {Flickr30k Entities: Collecting Region-to-Phrase Correspondences for Richer Image-to-Sentence Models - IEEE Conference Publication}, doi = {10.1109/ICCV.2015.303}, interhash = {0d506ec7cb5b56cf2437fc495a0df9f3}, intrahash = {d7874dca41004be97d7fc817fa9477c7}, issn = {2380-7504}, keywords = {flickr localisation nlp object paper-dzhi text}, month = dec, pages = {2641-2649}, timestamp = {2020-01-20T10:04:10.000+0100}, title = {Flickr30k Entities: Collecting Region-to-Phrase Correspondences for Richer Image-to-Sentence Models}, url = {https://ieeexplore.ieee.org/document/7410660}, year = 2015 }