Representing GPS trajectories as hyperspectral images enables multi-month dense anomaly detection that was previously computationally intractable.
March 27, 2026
Original Paper
Hyperspectral Trajectory Image for Multi-Month Trajectory Anomaly Detection
arXiv · 2603.25255
The Takeaway
By reframing trajectory analysis as a vision problem, TITAnD unifies sparse and dense GPS data into a single transformer-based framework. This allows practitioners to use mature CV architectures for urban mobility and fraud detection tasks at scale.
From the abstract
Trajectory anomaly detection underpins applications from fraud detection to urban mobility analysis. Dense GPS methods preserve fine-grained evidence such as abnormal speeds and short-duration events, but their quadratic cost makes multi-month analysis intractable; consequently, no existing approach detects anomalies over multi-month dense GPS trajectories. The field instead relies on scalable sparse stay-point methods that discard this evidence, forcing separate architectures for each regime an