Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Adaptive Human-centric Video Anomaly Detection: A Comprehensive Framework and A New Benchmark

Armin Danesh Pazho, Shanle Yao, Ghazal Alinezhad Noghre, Babak Rahimi Ardabili, Vinit Katariya, Hamed Tabkhi

TL;DR

The paper tackles the challenge of robust human-centric video anomaly detection in open, real-world settings by proposing HuVAD, the largest continuously recorded, privacy-enhanced dataset, and UCAL, an unsupervised continual anomaly learning framework that enables per-environment adaptation. It defines a standard HuVAD-S benchmark and introduces HuVAD-C for continual learning, demonstrating that UCAL-augmented models achieve state-of-the-art performance in a majority of cases and significantly improve adaptation over static training. The contributions include a rigorous privacy-preserving annotation pipeline, diverse real-world scenes, a comprehensive multi-metric evaluation (AUC-ROC, AUC-PR, EER, 10ER), and a first continual learning benchmark for human-centric VAD, with practical impact for deploying adaptive, privacy-conscious surveillance systems.

Abstract

Human-centric Video Anomaly Detection (VAD) aims to identify human behaviors that deviate from normal. At its core, human-centric VAD faces substantial challenges, such as the complexity of diverse human behaviors, the rarity of anomalies, and ethical constraints. These challenges limit access to high-quality datasets and highlight the need for a dataset and framework supporting continual learning. Moving towards adaptive human-centric VAD, we introduce the HuVAD (Human-centric privacy-enhanced Video Anomaly Detection) dataset and a novel Unsupervised Continual Anomaly Learning (UCAL) framework. UCAL enables incremental learning, allowing models to adapt over time, bridging traditional training and real-world deployment. HuVAD prioritizes privacy by providing de-identified annotations and includes seven indoor/outdoor scenes, offering over 5x more pose-annotated frames than previous datasets. Our standard and continual benchmarks, utilize a comprehensive set of metrics, demonstrating that UCAL-enhanced models achieve superior performance in 82.14% of cases, setting a new state-of-the-art (SOTA). The dataset can be accessed at https://github.com/TeCSAR-UNCC/HuVAD.

Towards Adaptive Human-centric Video Anomaly Detection: A Comprehensive Framework and A New Benchmark

TL;DR

The paper tackles the challenge of robust human-centric video anomaly detection in open, real-world settings by proposing HuVAD, the largest continuously recorded, privacy-enhanced dataset, and UCAL, an unsupervised continual anomaly learning framework that enables per-environment adaptation. It defines a standard HuVAD-S benchmark and introduces HuVAD-C for continual learning, demonstrating that UCAL-augmented models achieve state-of-the-art performance in a majority of cases and significantly improve adaptation over static training. The contributions include a rigorous privacy-preserving annotation pipeline, diverse real-world scenes, a comprehensive multi-metric evaluation (AUC-ROC, AUC-PR, EER, 10ER), and a first continual learning benchmark for human-centric VAD, with practical impact for deploying adaptive, privacy-conscious surveillance systems.

Abstract

Human-centric Video Anomaly Detection (VAD) aims to identify human behaviors that deviate from normal. At its core, human-centric VAD faces substantial challenges, such as the complexity of diverse human behaviors, the rarity of anomalies, and ethical constraints. These challenges limit access to high-quality datasets and highlight the need for a dataset and framework supporting continual learning. Moving towards adaptive human-centric VAD, we introduce the HuVAD (Human-centric privacy-enhanced Video Anomaly Detection) dataset and a novel Unsupervised Continual Anomaly Learning (UCAL) framework. UCAL enables incremental learning, allowing models to adapt over time, bridging traditional training and real-world deployment. HuVAD prioritizes privacy by providing de-identified annotations and includes seven indoor/outdoor scenes, offering over 5x more pose-annotated frames than previous datasets. Our standard and continual benchmarks, utilize a comprehensive set of metrics, demonstrating that UCAL-enhanced models achieve superior performance in 82.14% of cases, setting a new state-of-the-art (SOTA). The dataset can be accessed at https://github.com/TeCSAR-UNCC/HuVAD.
Paper Structure (15 sections, 7 figures, 3 tables, 1 algorithm)

This paper contains 15 sections, 7 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Works
  3. Video Anomaly Detection Dataset
  4. Continual Learning
  5. Privacy and Ethical Considerations
  6. Data Collection and Annotation
  7. Annotation Methodology
  8. Data Statistics
  9. Metrics
  10. HuVAD Standard Benchmarks (HuVAD-S)
  11. Standard Results
  12. Unsupervised Continual Anomaly Learning
  13. Methodology
  14. UCAL Results
  15. Conclusion

Figures (7)

  • Figure 1: Sample of anomalies and their annotations in the new proposed benchmark: HuVAD dataset. Cropped for visualization purposes. Segmentation is solely used for demonstration purposes.
  • Figure 2: The camera views excluding people. The ratio has been adjusted to fit the manuscript.
  • Figure 3: Samples from CSC Camera. Segmentation is solely used for demonstration purposes.
  • Figure 4: Pose counts per camera across major multi-camera datasets (SHT liu2018future, CHAD danesh2023chad, and NWPUC Cao_2023_CVPR).
  • Figure 5: IoU and crowd density across key datasets (SHT liu2018future, IITB rodrigues2020multi, CHAD danesh2023chad, and NWPUC Cao_2023_CVPR) and HuVAD's camera views.
  • ...and 2 more figures