Source-linked AI summary
FocusAdapt: Context-aware Adaptive Focus Assistance in Diminished Reality
Tianyu Zhang, Shutong Wu, Jiankun Yang, Zhen Bai, Yukang Yan
TL;DR
Diminished Reality can reduce clutter, but removing every task-irrelevant object may discard useful context and reduce situational awareness. FocusAdapt predicts object-level distraction from perceptual, semantic, and user-behavior information, then selectively suppresses distracting objects. A formative study found selective diminishing reduced cognitive load and improved task efficiency while preserving contextual information.
Problem
Diminished Reality needs to reduce visual clutter without removing task-irrelevant objects that provide useful context or situational awareness.
Method
FocusAdapt predicts distraction by combining visual saliency, similarity, semantic relevance, and user behavior to selectively suppress objects during procedural tasks.
Results
The formative study found that selective diminished reality reduced cognitive load and improved task efficiency while preserving useful contextual information.
Takeaways & Limitations
Diminished-reality systems can move beyond uniform object removal toward adaptive, personalized attention support.
Abstract
from arXiv · showhide
Diminished Reality (DR) can reduce visual clutter by removing irrelevant objects. However, removing all task-irrelevant objects may eliminate useful contextual information and reduce situational awareness. We present FocusAdapt, a context-aware DR system that predicts object-level distraction by integrating visual saliency, semantic relevance, and gaze behavior. Based on findings from a formative study, FocusAdapt selectively diminishes highly distracting objects while preserving useful context, enabling adaptive focus assistance during procedural tasks.
1 Introduction
FocusAdapt addresses the tension between reducing visual clutter and preserving useful context in procedural tasks. It selectively diminishes highly distracting objects using predicted attention informed by saliency, semantic relevance, and gaze-related findings.
- Removing all task-irrelevant objects may disrupt context, discard environmental cues, and reduce situational awareness.
- A formative study found selective diminishing reduced cognitive load as effectively as removing all task-irrelevant objects relative to no diminishing, across two tasks.
- Participants valued retaining some task-irrelevant objects as contextual cues, spatial anchors, and representations of personal relevance.
- FocusAdapt predicts distraction by integrating visual saliency analysis with task-aware semantic reasoning and selectively diminishes significant distractors.
2 User Study
The user study examined how visual clutter affects procedural-task performance and experience. Partial clutter removal was designed to remove gaze-selected distractors while retaining other environmental context.
- Twenty-four participants completed Room Organization and Block Assembly under high, low, and partial visual-clutter conditions.The within-subject study counterbalanced condition order across two sessions.
- Both low and partial clutter reduced perceived cognitive load, task completion time, and increased fixation on task-relevant objects.
- p < 0.01 for mental demand, while completion time decreased versus high clutter in room arrangement (p < 0.01) and block assembly (p < 0.001).
- Low clutter increased fixation ratios in selected room-arrangement and block-assembly subtasks, while partial clutter increased the ratio in room-arrangement subtask 6.
- Participants preferred diminished-reality systems that were selective, context-aware, adaptive, and user-controllable rather than overly aggressive.
3 FocusAdapt
FocusAdapt is a selective diminished-reality system for procedural tasks. Its pipeline combines scene analysis, perceptual and semantic features, and attention prediction to identify objects for suppression.
- FocusAdapt was developed to assist users completing procedural tasks through selective diminished reality.
- The system analyzes objects using detection, segmentation, feature embeddings, visual saliency, and task-dependent semantic relevance.
- An attention model combines bottom-up perceptual features with top-down features from the current task step, object category, and task relevance.
- The prototype precomputed computationally intensive scene-understanding components and replayed them during the study to approximate real-time behavior.
4 Conclusion
FocusAdapt selectively suppresses distracting objects by integrating perceptual, semantic, and user-behavior information. The study supports adaptive attention assistance that preserves useful context during procedural tasks.
- FocusAdapt integrates visual saliency and similarity, semantic relevance, and user behavior to selectively suppress distracting objects.
- The formative study found that selective diminished reality reduces cognitive load, improves task efficiency, and preserves useful contextual information.
- The findings support moving beyond one-size-fits-all object removal toward adaptive, personalized attention support.