Similar Items: From Model Uncertainty to Human Attention: Localization-Aware Visual Cues for Scalable Annotation Review