PulseAugur
LIVE 12:22:17
research · [2 sources] ·
0
research

MSD-Score metric improves image caption evaluation without references

By PulseAugur Editorial · Summary by gemini-2.5-flash-lite from 2 sources

Researchers have developed MSD-Score, a novel method for evaluating image captions without needing reference captions. This approach models image patch and text token embeddings as distributions, enabling a more nuanced assessment of semantic discrepancies. MSD-Score achieves state-of-the-art correlation with human judgments and offers transparent diagnostics for local grounding errors. AI

Summary written by gemini-2.5-flash-lite from 2 sources. How we write summaries →

IMPACT Introduces a new reference-free metric for image caption evaluation that correlates highly with human judgment.

RANK_REASON The cluster contains an academic paper detailing a new evaluation metric for image captions.

Read on arXiv cs.CV →

COVERAGE [2]

  1. arXiv cs.CV TIER_1 · Shichao Kan, Xuyang Zhang, Haojie Zhang, Zhe Zhu, Yigang Cen, Yixiong Liang, Lianlei Shan, Linna Zhang, Zhe Qu, Jiazhi Xia ·

    MSD-Score: Multi-Scale Distributional Scoring for Reference-Free Image Caption Evaluation

    arXiv:2605.06080v1 Announce Type: new Abstract: Evaluating image captions without references remains challenging because global embedding similarity often misses fine-grained mismatches such as hallucinated objects, missing attributes, or incorrect relations. We propose MSD-Score…

  2. arXiv cs.CV TIER_1 · Jiazhi Xia ·

    MSD-Score: Multi-Scale Distributional Scoring for Reference-Free Image Caption Evaluation

    Evaluating image captions without references remains challenging because global embedding similarity often misses fine-grained mismatches such as hallucinated objects, missing attributes, or incorrect relations. We propose MSD-Score, a reference-free metric that models image patc…

RELATED ENTITIES

RELATED TOPICS