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

TL;DR

This paper demonstrates that visualizing localization uncertainty in AI-assisted annotation workflows improves label quality and efficiency by guiding annotators to focus on uncertain predictions.

Contribution

It introduces a novel visualization of spatial uncertainty and shows its effectiveness in enhancing human-in-the-loop annotation accuracy and speed.

Findings

Uncertainty cues lead to higher label quality.

Annotators focus more on uncertain predictions.

Overall annotation speed improves with uncertainty visualization.

Abstract

High-quality labeled data is essential for training robust machine learning models, yet obtaining annotations at scale remains expensive. AI-assisted annotation has therefore become standard in large-scale labeling workflows. However, in tasks where model predictions carry two independent components, a class label and spatial boundaries, a model may classify an object with high confidence while mislocalizing it. Existing AI-assisted workflows offer annotators no signal about where spatial errors are most likely. Without such guidance, humans may systematically underinspect subtly misplaced boxes. We address this by studying the effect of visualizing spatial uncertainty via a purpose-built interface. In a controlled study with 120 participants, those receiving uncertainty cues achieve higher label quality while being faster overall. A box-level analysis confirms that the cues redirect annotator effort toward high-uncertainty predictions and away from well-localized boxes. These findings establish localization uncertainty as a lever to improve human-in-the-loop annotation. Code is available at https://mos-ks.github.io/MUHA/.