This paper shows that CIDEr-D, a traditional evaluation metric for image description, does not work properly on datasets where the number of words in the sentence is significantly greater than those in the MS COCO Captions dataset. We also show that CIDEr-D has performance hampered by the lack of multiple reference sentences and high variance of sentence length. To bypass this problem, we introduce CIDEr-R, which improves CIDEr-D, making it more flexible in dealing with datasets with high sentence length variance. We demonstrate that CIDEr-R is more accurate and closer to human judgment than CIDEr-D; CIDEr-R is more robust regarding the number of available references. Our results reveal that using Self-Critical Sequence Training to optimize CIDEr-R generates descriptive captions. In contrast, when CIDEr-D is optimized, the generated captions' length tends to be similar to the reference length. However, the models also repeat several times the same word to increase the sentence length.
Automatically describing images using natural sentences is essential to visually impaired people’s inclusion on the Internet. This problem is known as Image Captioning. There are many datasets in the literature, but most contain only English captions, whereas datasets with captions described in other languages are scarce. We introduce the #PraCegoVer, a multi-modal dataset with Portuguese captions based on posts from Instagram. It is the first large dataset for image captioning in Portuguese. In contrast to popular datasets, #PraCegoVer has only one reference per image, and both mean and variance of reference sentence length are significantly high, which makes our dataset challenging due to its linguistic aspect. We carry a detailed analysis to find the main classes and topics in our data. We compare #PraCegoVer to MS COCO dataset in terms of sentence length and word frequency. We hope that #PraCegoVer dataset encourages more works addressing the automatic generation of descriptions in Portuguese.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.