Revisiting Image-Language Networks for Open-Ended Phrase Detection

Most existing work that grounds natural language phrases in images starts with the assumption that the phrase in question is relevant to the image. In this paper we address a more realistic version of the natural language grounding task where we must both identify whether the phrase is relevant to a...

Ausführliche Beschreibung

Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on pattern analysis and machine intelligence. - 1979. - 44(2022), 4 vom: 06. Apr., Seite 2155-2167
1. Verfasser: Plummer, Bryan A (VerfasserIn)
Weitere Verfasser: Shih, Kevin J, Li, Yichen, Xu, Ke, Lazebnik, Svetlana, Sclaroff, Stan, Saenko, Kate
Format: Online-Aufsatz
Sprache:English
Veröffentlicht: 2022
Zugriff auf das übergeordnete Werk:IEEE transactions on pattern analysis and machine intelligence
Schlagworte:Journal Article Research Support, U.S. Gov't, Non-P.H.S.
LEADER 01000caa a22002652c 4500
001 NLM315921331
003 DE-627
005 20250228041138.0
007 cr uuu---uuuuu
008 231225s2022 xx |||||o 00| ||eng c
024 7 |a 10.1109/TPAMI.2020.3029008  |2 doi 
028 5 2 |a pubmed25n1052.xml 
035 |a (DE-627)NLM315921331 
035 |a (NLM)33021939 
040 |a DE-627  |b ger  |c DE-627  |e rakwb 
041 |a eng 
100 1 |a Plummer, Bryan A  |e verfasserin  |4 aut 
245 1 0 |a Revisiting Image-Language Networks for Open-Ended Phrase Detection 
264 1 |c 2022 
336 |a Text  |b txt  |2 rdacontent 
337 |a ƒaComputermedien  |b c  |2 rdamedia 
338 |a ƒa Online-Ressource  |b cr  |2 rdacarrier 
500 |a Date Completed 28.03.2022 
500 |a Date Revised 01.04.2022 
500 |a published: Print-Electronic 
500 |a Citation Status MEDLINE 
520 |a Most existing work that grounds natural language phrases in images starts with the assumption that the phrase in question is relevant to the image. In this paper we address a more realistic version of the natural language grounding task where we must both identify whether the phrase is relevant to an image and localize the phrase. This can also be viewed as a generalization of object detection to an open-ended vocabulary, introducing elements of few- and zero-shot detection. We propose an approach for this task that extends Faster R-CNN to relate image regions and phrases. By carefully initializing the classification layers of our network using canonical correlation analysis (CCA), we encourage a solution that is more discerning when reasoning between similar phrases, resulting in over double the performance compared to a naive adaptation on three popular phrase grounding datasets, Flickr30K Entities, ReferIt Game, and Visual Genome, with test-time phrase vocabulary sizes of 5K, 32K, and 159K, respectively 
650 4 |a Journal Article 
650 4 |a Research Support, U.S. Gov't, Non-P.H.S. 
700 1 |a Shih, Kevin J  |e verfasserin  |4 aut 
700 1 |a Li, Yichen  |e verfasserin  |4 aut 
700 1 |a Xu, Ke  |e verfasserin  |4 aut 
700 1 |a Lazebnik, Svetlana  |e verfasserin  |4 aut 
700 1 |a Sclaroff, Stan  |e verfasserin  |4 aut 
700 1 |a Saenko, Kate  |e verfasserin  |4 aut 
773 0 8 |i Enthalten in  |t IEEE transactions on pattern analysis and machine intelligence  |d 1979  |g 44(2022), 4 vom: 06. Apr., Seite 2155-2167  |w (DE-627)NLM098212257  |x 1939-3539  |7 nnas 
773 1 8 |g volume:44  |g year:2022  |g number:4  |g day:06  |g month:04  |g pages:2155-2167 
856 4 0 |u http://dx.doi.org/10.1109/TPAMI.2020.3029008  |3 Volltext 
912 |a GBV_USEFLAG_A 
912 |a SYSFLAG_A 
912 |a GBV_NLM 
912 |a GBV_ILN_350 
951 |a AR 
952 |d 44  |j 2022  |e 4  |b 06  |c 04  |h 2155-2167