Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Advancing Video Anomaly Detection: A Concise Review and a New Dataset

Liyun Zhu, Lei Wang, Arjun Raj, Tom Gedeon, Chen Chen

TL;DR

A new dataset, Multi-Scenario Anomaly Detection (MSAD), comprising 14 distinct scenarios captured from various camera views is introduced, providing a robust foundation for training superior models.

Abstract

Video Anomaly Detection (VAD) finds widespread applications in security surveillance, traffic monitoring, industrial monitoring, and healthcare. Despite extensive research efforts, there remains a lack of concise reviews that provide insightful guidance for researchers. Such reviews would serve as quick references to grasp current challenges, research trends, and future directions. In this paper, we present such a review, examining models and datasets from various perspectives. We emphasize the critical relationship between model and dataset, where the quality and diversity of datasets profoundly influence model performance, and dataset development adapts to the evolving needs of emerging approaches. Our review identifies practical issues, including the absence of comprehensive datasets with diverse scenarios. To address this, we introduce a new dataset, Multi-Scenario Anomaly Detection (MSAD), comprising 14 distinct scenarios captured from various camera views. Our dataset has diverse motion patterns and challenging variations, such as different lighting and weather conditions, providing a robust foundation for training superior models. We conduct an in-depth analysis of recent representative models using MSAD and highlight its potential in addressing the challenges of detecting anomalies across diverse and evolving surveillance scenarios. [Project website: https://msad-dataset.github.io/]

Advancing Video Anomaly Detection: A Concise Review and a New Dataset

TL;DR

A new dataset, Multi-Scenario Anomaly Detection (MSAD), comprising 14 distinct scenarios captured from various camera views is introduced, providing a robust foundation for training superior models.

Abstract

Video Anomaly Detection (VAD) finds widespread applications in security surveillance, traffic monitoring, industrial monitoring, and healthcare. Despite extensive research efforts, there remains a lack of concise reviews that provide insightful guidance for researchers. Such reviews would serve as quick references to grasp current challenges, research trends, and future directions. In this paper, we present such a review, examining models and datasets from various perspectives. We emphasize the critical relationship between model and dataset, where the quality and diversity of datasets profoundly influence model performance, and dataset development adapts to the evolving needs of emerging approaches. Our review identifies practical issues, including the absence of comprehensive datasets with diverse scenarios. To address this, we introduce a new dataset, Multi-Scenario Anomaly Detection (MSAD), comprising 14 distinct scenarios captured from various camera views. Our dataset has diverse motion patterns and challenging variations, such as different lighting and weather conditions, providing a robust foundation for training superior models. We conduct an in-depth analysis of recent representative models using MSAD and highlight its potential in addressing the challenges of detecting anomalies across diverse and evolving surveillance scenarios. [Project website: https://msad-dataset.github.io/]
Paper Structure (19 sections, 5 equations, 7 figures, 8 tables)

This paper contains 19 sections, 5 equations, 7 figures, 8 tables.

Table of Contents

  1. Introduction
  2. A Concise Review
  3. Dataset deficiencies and biases
  4. Discussion on model and dataset evolution
  5. A Multi-Scenario Dataset
  6. Experiment
  7. Generalizability and adaptability
  8. Practical applicability and effectiveness
  9. Limitations of current methods on our new dataset
  10. Conclusion
  11. Dataset collection process
  12. Further dataset details
  13. Dataset usage and maintenance
  14. Scenario-adaptive anomaly detection framework
  15. Further discussions
  16. ...and 4 more sections

Figures (7)

  • Figure 1: A comparison of existing datasets such as UCSD Ped, CUHK Avenue, ShanghaiTech, UCF-Crime, UBnormal and CUVA vs. our Multi-Scenario Anomaly Detection (MSAD) dataset.
  • Figure 2: Our MSAD includes a diverse range of scenarios, both indoor and outdoor, featuring various objects, e.g., pedestrians, cars, trains, etc. The first row shows different real-world common motions, while the second row demonstrates variations in weather and lighting conditions. The third row displays different moving objects. The last column shows human- and non-human-related anomalies.
  • Figure 3: The statistics of our MSAD dataset include: (a) a breakdown of main anomaly types and their respective percentages, (b) a boxplot illustrating frame number variations across scenarios in MSAD training set, and (c) the distributions of train/test splits across scenarios for two evaluation protocols (see Sec. \ref{['sec: dataset']} evaluation protocols): (top plot) generalizability and adaptability, and (bottom plot) practical applicability and effectiveness.
  • Figure 4: Distributions of (a) frame number variations, (b) video resolutions, (c) video durations, and (d) detailed distributions per anomaly type per scenario in the entire MSAD dataset are presented. Our main paper presents the distribution of frame number variations in the training set (see Fig. \ref{['subfig-dis']}).
  • Figure 5: Visualizations of anomaly frames and their corresponding mean motion maps. The mean motion map is calculated by averaging the frame difference maps obtained from pairs of consecutive frames.
  • ...and 2 more figures