JaSPICE: Automatic Evaluation Metric Using Predicate-Argument Structures for Image Captioning Models
This work addresses the problem of evaluating image captioning models in Japanese for researchers, but it is incremental as it adapts an existing English metric to a new language.
The authors tackled the lack of automatic evaluation metrics for Japanese image captioning by proposing JaSPICE, a metric based on scene graphs, which outperformed baseline metrics in correlation with human evaluations on a dataset of 103,170 human judgments.
Image captioning studies heavily rely on automatic evaluation metrics such as BLEU and METEOR. However, such n-gram-based metrics have been shown to correlate poorly with human evaluation, leading to the proposal of alternative metrics such as SPICE for English; however, no equivalent metrics have been established for other languages. Therefore, in this study, we propose an automatic evaluation metric called JaSPICE, which evaluates Japanese captions based on scene graphs. The proposed method generates a scene graph from dependencies and the predicate-argument structure, and extends the graph using synonyms. We conducted experiments employing 10 image captioning models trained on STAIR Captions and PFN-PIC and constructed the Shichimi dataset, which contains 103,170 human evaluations. The results showed that our metric outperformed the baseline metrics for the correlation coefficient with the human evaluation.