# From Frames to Events: Rethinking Evaluation in Human-Centric Video Anomaly Detection

Source: [From Frames to Events: Rethinking Evaluation in Human-Centric Video Anomaly Detection](<https://www.sota2.com/research/paper/69dc78767d0172140ec78a4e>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-14T05:00:11.146Z

Authors: Narges Rashvand, Shanle Yao, Armin Danesh Pazho, Babak Rahimi Ardabili, Hamed Tabkhi

Published: 2026-04-10T13:52:18.000Z

## Abstract

Pose-based Video Anomaly Detection (VAD) has gained significant attention for its privacy-preserving nature and robustness to environmental variations. However, traditional frame-level evaluations treat video as a collection of isolated frames, fundamentally misaligned with how anomalies manifest and are acted upon in the real world. In operational surveillance systems, what matters is not the flagging of individual frames, but the reliable detection, localization, and reporting of a coherent anomalous event, a contiguous temporal episode with an identifiable onset and duration. Frame-level metrics are blind to this distinction, and as a result, they systematically overestimate model performance for any deployment that requires actionable, event-level alerts. In this work, we propose a shift toward an event-centric perspective in VAD. We first audit widely used VAD benchmarks, including SHT\[19\], CHAD\[6\], NWPUC\[4\], and HuVAD\[25\], to characterize their event structure. We then introduce two strategies for temporal event localization: a score-refinement pipeline with hierarchical Gaussian smoothing and adaptive binarization, and an end-to-end Dual-Branch Model that directly generates event-level detections. Finally, we establish the first event-based evaluation standard for VAD by adapting Temporal Action Localization metrics, including tIoU-based event matching and multi-threshold F1 evaluation. Our results quantify a substantial performance gap: while all SoTA models achieve frame-level AUC-ROC exceeding 52% on the NWPUC\[4\], their event-level localization precision falls below 10% even at a minimal tIoU=0.2, with an average event-level F1 of only 0.11 across all thresholds. The code base for this work is available at https://github.com/TeCSAR-UNCC/EventCentric-VAD.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Event-level Video Anomaly Detection on CHAD 6 (test)](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-chad-6-test.md>): Event-level Video Anomaly Detection; CHAD 6 (test); tIOU=0.5 Precision: 16.71.

- [Event-level Video Anomaly Detection on HuVAD 25 (test)](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-huvad-25-test.md>): Event-level Video Anomaly Detection; HuVAD (Human Video Anomaly Detection) 25 (test); Precision (tIOU=0.5): 22.2.

- [Video Anomaly Detection on ShanghaiTech (SHT) (tIOU-based F1)](<https://www.sota2.com/research/sota/video-anomaly-detection-on-shanghaitech-sht-tiou-based-f1.md>): Video Anomaly Detection; SHT (ShanghaiTech); Precision (tIOU=0.5): 56.58.

- [Video Anomaly Detection on CHAD (tIOU-based F1 scores)](<https://www.sota2.com/research/sota/video-anomaly-detection-on-chad-tiou-based-f1-scores.md>): Video Anomaly Detection; CHAD; Precision (tIOU=0.5): 17.29.

- [Video Anomaly Detection on NWPUC (NWPU Campus)](<https://www.sota2.com/research/sota/video-anomaly-detection-on-nwpuc-nwpu-campus.md>): Video Anomaly Detection; NWPUC (NWPU Campus); Precision (tIOU=0.5): 4.15.

- [Video Anomaly Detection on HuVAD](<https://www.sota2.com/research/sota/video-anomaly-detection-on-huvad.md>): Video Anomaly Detection; HuVAD; Precision (tIOU=0.5): 22.77.

- [Event-level Video Anomaly Detection on ShanghaiTech (SHT) 19 (test)](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-shanghaitech-sht-19-test.md>): Event-level Video Anomaly Detection; ShanghaiTech (SHT) 19 (test); Precision @ tIOU=0.5: 46.27.

- [Event-level Video Anomaly Detection on NWPU Campus (NWPUC) 4 (test)](<https://www.sota2.com/research/sota/event-level-video-anomaly-detection-on-nwpu-campus-nwpuc-4-test.md>): Event-level Video Anomaly Detection; NWPU Campus (NWPUC) 4 (test); Precision (tIOU=0.5): 4.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.09327>)

[PDF](<https://arxiv.org/pdf/2604.09327.pdf>)
