"EveLoad: Cognitive Workload Recognition from Eye Movements Using Event Cameras"

"Cognitive workload plays a critical role in immersive and interactive systems such as augmented and virtual reality (AR\/VR). Accurate awareness of cognitive workload helps enable adaptive rendering, interaction, and content delivery. Existing eye-movement-based cognitive workload recognition methods mainly rely on frame-based conventional eye trackers, which often suffer from limited temporal resolution and degraded robustness under rapid eye movements. In contrast, event cameras provide microsecond-level temporal resolution and high dynamic range, enabling reliable capture of fine-grained ocular dynamics. Moreover, most prior studies adopt free-viewing paradigms, where spatial gaze patterns dominate workload discrimination, leaving the problem of cognitive workload recognition under task-constrained conditions largely underexplored.In this work, to the best of our knowledge, we introduce EveLoad, the first event-based eye-movement dataset with graded cognitive workload annotations, collected from 15 subjects under spatially constrained and task-driven conditions using a controlled N-back-guided fixation paradigm. Leveraging this dataset, we establish a benchmark for cognitive workload recognition with six progressively increasing workload levels and propose a learning framework that encodes spatiotemporal event representations. Experimental results show that the proposed method achieves an average subject-specific accuracy of 95.5% and 94.3% under random split evaluation, highlighting the potential of event cameras for cognitive state monitoring in immersive and interactive systems."