Guided Open Vocabulary Image Captioning with Constrained Beam Search
This addresses the limitation of existing captioning models in real-world applications with diverse images, representing an incremental improvement by enhancing generalization without architectural changes.
The paper tackles the problem of image captioning models failing to generalize to out-of-domain images by introducing a method that uses constrained beam search to incorporate tag words from image taggers without retraining, achieving state-of-the-art results on MSCOCO for out-of-domain captioning and improved in-domain performance.
Existing 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.