Contrastive Semantic Similarity Learning for Image Captioning Evaluation with Intrinsic Auto-encoder
This work addresses the problem of flexible language expression in image captioning evaluation for researchers and practitioners, offering a complementary metric that is incremental in nature.
The paper tackles the challenge of evaluating image caption quality by proposing a learning-based metric, I^2CE, which uses contrastive learning and auto-encoders to capture sentence-level information, achieving better consistency with human judgments compared to contemporary metrics.
Automatically evaluating the quality of image captions can be very challenging since human language is quite flexible that there can be various expressions for the same meaning. Most of the current captioning metrics rely on token level matching between candidate caption and the ground truth label sentences. It usually neglects the sentence-level information. Motivated by the auto-encoder mechanism and contrastive representation learning advances, we propose a learning-based metric for image captioning, which we call Intrinsic Image Captioning Evaluation($I^2CE$). We develop three progressive model structures to learn the sentence level representations--single branch model, dual branches model, and triple branches model. Our empirical tests show that $I^2CE$ trained with dual branches structure achieves better consistency with human judgments to contemporary image captioning evaluation metrics. Furthermore, We select several state-of-the-art image captioning models and test their performances on the MS COCO dataset concerning both contemporary metrics and the proposed $I^2CE$. Experiment results show that our proposed method can align well with the scores generated from other contemporary metrics. On this concern, the proposed metric could serve as a novel indicator of the intrinsic information between captions, which may be complementary to the existing ones.