MMVP

该数据集旨在通过配对分类准确率来评估对不同视觉模式中视觉细节的理解能力。它需要使用线性混合模型（LMMs）来捕捉微妙的视觉差异。该数据集涵盖了9种不同的视觉模式，其任务是深入理解视觉细节。

This dataset is designed to evaluate the comprehension of visual details across different visual modalities via paired classification accuracy. It requires the use of Linear Mixed Models (LMMs) to capture subtle visual differences. This dataset covers 9 distinct visual modalities, whose tasks focus on in-depth understanding of visual details.