Back to Search Start Over

Exploring Nearest Neighbor Approaches for Image Captioning

Authors :
Devlin, Jacob
Gupta, Saurabh
Girshick, Ross
Mitchell, Margaret
Zitnick, C. Lawrence
Publication Year :
2015

Abstract

We explore a variety of nearest neighbor baseline approaches for image captioning. These approaches find a set of nearest neighbor images in the training set from which a caption may be borrowed for the query image. We select a caption for the query image by finding the caption that best represents the "consensus" of the set of candidate captions gathered from the nearest neighbor images. When measured by automatic evaluation metrics on the MS COCO caption evaluation server, these approaches perform as well as many recent approaches that generate novel captions. However, human studies show that a method that generates novel captions is still preferred over the nearest neighbor approach.

Details

Database :
arXiv
Publication Type :
Report
Accession number :
edsarx.1505.04467
Document Type :
Working Paper