SPICE: Semantic Propositional Image Caption Evaluation
arXiv:1607.08822
Abstract
There is considerable interest in the task of automatically generating image captions. However, evaluation is challenging. Existing automatic evaluation metrics are primarily sensitive to n-gram overlap, which is neither necessary nor sufficient for the task of simulating human judgment. We hypothesize that semantic propositional content is an important component of human caption evaluation, and propose a new automated caption evaluation metric defined over scene graphs coined SPICE. Extensive evaluations across a range of models and datasets indicate that SPICE captures human judgments over model-generated captions better than other automatic metrics (e.g., system-level correlation of 0.88 with human judgments on the MS COCO dataset, versus 0.43 for CIDEr and 0.53 for METEOR). Furthermore, SPICE can answer questions such as `which caption-generator best understands colors?' and `can caption-generators count?'
14 pages plus references, accepted to ECCV 2016
References in corpus (8)
- Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
- Microsoft COCO Captions: Data Collection and Evaluation Server
- Show and Tell: A Neural Image Caption Generator
- Exploring Nearest Neighbor Approaches for Image Captioning
- Language Models for Image Captioning: The Quirks and What Works
- Flickr30k Entities: Collecting Region-to-Phrase Correspondences for Richer Image-to-Sentence Models
- From Images to Sentences through Scene Description Graphs using Commonsense Reasoning and Knowledge
- Technical Report: Image Captioning with Semantically Similar Images