Anderson, PeterFernando, BasuraJohnson, MarkGould, StephenMartha PalmerRebecca HwaSebastian Riedel2023-07-202023-07-20September978-1-945626-83-8http://hdl.handle.net/1885/294445Existing image captioning models do not generalize well to out-of-domain images containing novel scenes or objects. This limitation severely hinders the use of these models in real world applications dealing with images in the wild. We address this problem using a flexible approach that enables existing deep captioning architectures to take advantage of image taggers at test time, without re-training. Our method uses constrained beam search to force the inclusion of selected tag words in the output, and fixed, pretrained word embeddings to facilitate vocabulary expansion to previously unseen tag words. Using this approach we achieve state of the art results for out-of-domain captioning on MSCOCO (and improved results for in-domain captioning). Perhaps surprisingly, our results significantly outperform approaches that incorporate the same tag predictions into the learning algorithm. We also show that we can significantly improve the quality of generated ImageNet captions by leveraging ground-truth labels.This research is supported by an Australian Government Research Training Program (RTP) Scholarship and by the Australian Research Council Centre of Excellence for Robotic Vision (project number CE140100016).application/pdfen-AU© 2017 Association for Computational Linguisticshttps://creativecommons.org/licenses/by/4.0/201710.18653/v1/D17-10982022-05-22Creative Commons Attribution 4.0 International License