mlsanityData to 14 Oct 2026

Bounding-Box Trajectories Matter for Video Anomaly Detection

arXiv 2605.21957 ↗First posted 21 May 2026cs.CVcited by 3

Not on the map. Why?

Abstract (arXiv)

Video anomaly detection is critical for public safety and security, yet remains highly challenging despite extensive research due to large variations in appearance, viewpoint, and scene dynamics. Among existing approaches, human pose-based methods have emerged as a major line of research, showing strong performance since many anomalies in public datasets involve humans and pose representations are robust to appearance changes while providing compact motion descriptions. However, these methods often overlook bounding-box trajectories, although such information is inherently available in pose-based pipelines. In this paper, we explicitly leverage these trajectories as a primary anomaly cue. We present TrajVAD, a framework that models multi-class bounding-box trajectories using normalizing flows to learn normal kinematic patterns. Its trajectory-only variant, TrajVAD-T, eliminates pose estimation, reaches 87.7 AP on ShanghaiTech, and achieves the best results on MSAD among compared methods. TrajVAD-P adds a reliability-gated pose branch and improves performance to 88.6 AUROC and 90.9 AP on ShanghaiTech, establishing bounding-box trajectories as an effective modality for video anomaly detection.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 39 of its 49 references that are in the map. First posted on arXiv 21 May 2026.

Builds on

39 of 49 references.

References · 39

  1. 2016Density estimation using Real NVPcited by 2,042
  2. 2018Glow: Generative Flow with Invertible 1x1 Convolutionscited by 1,545
  3. 2021YOLOX: Exceeding YOLO Series in 2021cited by 1,071
  4. 2021ByteTrack: Multi-Object Tracking by Associating Every Detection Boxcited by 823
  5. 2019Memorizing Normality to Detect Anomaly: Memory-augmented Deep Autoencoder for Unsupervised Anomaly Detectioncited by 564
  6. 2017Future Frame Prediction for Anomaly Detection -- A New Baselinecited by 391
  7. 2019Omni-Scale Feature Learning for Person Re-Identificationcited by 295
  8. 2016Learning Temporal Regularity in Video Sequencescited by 254
  9. 2022AlphaPose: Whole-Body Regional Multi-Person Pose Estimation and Tracking in Real-Timecited by 200
  10. 2018Latent Space Autoregression for Novelty Detectioncited by 149
  11. 2020Anomaly Detection in Video via Self-Supervised and Multi-Task Learningcited by 148
  12. 2021A Hybrid Video Anomaly Detection Framework via Memory-Augmented Flow Reconstruction and Flow-Guided Frame Predictioncited by 129
27 more
  1. 2019Learning Regularity in Skeleton Trajectories for Anomaly Detection in Videoscited by 104
  2. 2022Generative Cooperative Learning for Unsupervised Video Anomaly Detectioncited by 102
  3. 2019Graph Embedded Pose Clustering for Anomaly Detectioncited by 95
  4. 2021UBnormal: New Benchmark for Supervised Open-Set Video Anomaly Detectioncited by 93
  5. 2020A Background-Agnostic Framework with Adversarial Training for Abnormal Event Detection in Videocited by 86
  6. 2020Cloze Test Helps: Effective Video Anomaly Detection via Learning to Complete Video Eventscited by 70
  7. 2022Video Anomaly Detection by Solving Decoupled Spatio-Temporal Jigsaw Puzzlescited by 62
  8. 2022Normalizing Flows for Human Pose Anomaly Detectioncited by 60
  9. 2019Multi-timescale Trajectory Prediction for Abnormal Human Activity Detectioncited by 59
  10. 2017Joint Detection and Recounting of Abnormal Events by Learning Deep Generic Knowledgecited by 55
  11. 2024Advancing Video Anomaly Detection: A Concise Review and a New Datasetcited by 49
  12. 2017Unmasking the abnormal events in videocited by 48
  13. 2022SSMTL++: Revisiting Self-Supervised Multi-Task Learning for Video Anomaly Detectioncited by 45
  14. 2023Multimodal Motion Conditioned Diffusion Model for Skeleton-based Video Anomaly Detectioncited by 42
  15. 2024MULDE: Multiscale Log-Density Estimation via Denoising Score Matching for Video Anomaly Detectioncited by 22
  16. 2022EVAL: Explainable Video Anomaly Localizationcited by 19
  17. 2024An Exploratory Study on Human-Centric Video Anomaly Detection through Variational Autoencoders and Trajectory Predictioncited by 19
  18. 2023Holistic Representation Learning for Multitask Trajectory Anomaly Detectioncited by 15
  19. 2023Contracting Skeletal Kinematics for Human-Related Video Anomaly Detectioncited by 14
  20. 2024Graph-Jigsaw Conditioned Diffusion Model for Skeleton-based Video Anomaly Detectioncited by 9
  21. 2024Video Question Answering for People with Visual Impairments Using an Egocentric 360-Degree Cameracited by 9
  22. 2025Real-time Traffic Accident Anticipation with Feature Reusecited by 8
  23. 2022BiPOCO: Bi-Directional Trajectory Prediction with Pose Constraints for Pedestrian Anomaly Detectioncited by 7
  24. 2024DA-Flow: Dual Attention Normalizing Flow for Skeleton-based Video Anomaly Detectioncited by 7
  25. 2024Anomaly Detection for People with Visual Impairments Using an Egocentric 360-Degree Cameracited by 7
  26. 2025Sequential keypoint density estimator: an overlooked baseline of skeleton-based video anomaly detectioncited by 7
  27. 2026Instance-Aligned Captions for Explainable Video Anomaly Detectioncited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026Rethinking Open-World Video Anomaly Detection: Diagnosing Definition Blindnesscited by 2
  2. 2026A VLM Answer Is Not an Anomaly Score: Rank Compression Across Image and Video Anomaly Detectioncited by 1
  3. 2026Look Inside Each Video: Rethinking How Video Anomaly Detection Is Evaluatedcited by 0