CLIPScore: A Reference-free Evaluation Metric for Image Captioning
About
Image captioning has conventionally relied on reference-based automatic evaluations, where machine captions are compared against captions written by humans. This is in contrast to the reference-free manner in which humans assess caption quality. In this paper, we report the surprising empirical finding that CLIP (Radford et al., 2021), a cross-modal model pretrained on 400M image+caption pairs from the web, can be used for robust automatic evaluation of image captioning without the need for references. Experiments spanning several corpora demonstrate that our new reference-free metric, CLIPScore, achieves the highest correlation with human judgements, outperforming existing reference-based metrics like CIDEr and SPICE. Information gain experiments demonstrate that CLIPScore, with its tight focus on image-text compatibility, is complementary to existing reference-based metrics that emphasize text-text similarities. Thus, we also present a reference-augmented version, RefCLIPScore, which achieves even higher correlation. Beyond literal description tasks, several case studies reveal domains where CLIPScore performs well (clip-art images, alt-text rating), but also where it is relatively weaker in comparison to reference-based metrics, e.g., news captions that require richer contextual knowledge.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Object Hallucination Evaluation | POPE | Accuracy74.57 | 2056 | |
| Visual Question Answering | VizWiz | Accuracy43 | 1863 | |
| Text-based Visual Question Answering | TextVQA | Accuracy54.7 | 984 | |
| Image Classification | ImageNet 1k (test) | Top-1 Accuracy27.3 | 939 | |
| Science Question Answering | ScienceQA | Accuracy69.06 | 916 | |
| Multimodal Understanding | SEED-Bench | Accuracy61.7 | 571 | |
| Visual Question Answering | VQA v2 | Accuracy73.4 | 347 | |
| Scientific Question Answering | ScienceQA image | Accuracy65 | 281 | |
| Multi-modal Evaluation | MME | MME Score1.57e+3 | 240 | |
| Image Quality Assessment | AGIQA-3K | SRCC0.5207 | 175 |