Ego-centric Traffic Anomaly Detection (TAD) aims to identify abnormal events in videos captured by dashboard-mounted cameras in vehicles. Compared to anomaly detection in roadside surveillance videos, ego-centric TAD poses greater challenges due to the dynamic backgrounds caused by vehicle motion. Previous frame-level methods are often vulnerable to interference from these dynamic backgrounds and struggle to detect small objects located at a distance or off-center. To address these challenges, we propose an object-prompt enhanced method that integrates detected traffic objects into a frame-level TAD framework. Our approach introduces an object-prompt scheme comprising an object prompt encoder, along with two cross-attention-based aggregation modules: an instance-wise aggregation module for fusing information between object instances and the scene, and a relation-wise aggregation module for capturing relationships inter-objects. Additionally, we design an instance-level loss to supervise anomaly detection at the object level. Our method effectively mitigates interference from dynamic backgrounds, improves the detection of distant or off-center anomalies, and enables precise spatial localization of anomalies. Experimental results on the DoTA and DADA-2000 datasets demonstrate that our method achieves state-of-the-art performance.
The ability to understand the surrounding scene is of paramount importance for Autonomous Vehicles (AVs). This paper presents a system capable to work in an online fashion, giving an immediate response to the arise of anomalies surrounding the AV, exploiting only the videos captured by a dash-mounted camera. Our architecture, called MOVAD, relies on two main modules: a Short-Term Memory Module to extract information related to the ongoing action, implemented by a Video Swin Transformer (VST), and a Long-Term Memory Module injected inside the classifier that considers also remote past information and action context thanks to the use of a Long-Short Term Memory (LSTM) network. The strengths of MOVAD are not only linked to its excellent performance, but also to its straightforward and modular architecture, trained in a end-to-end fashion with only RGB frames with as less assumptions as possible, which makes it easy to implement and play with. We evaluated the performance of our method on Detection of Traffic Anomaly (DoTA) dataset, a challenging collection of dash-mounted camera videos of accidents. After an extensive ablation study, MOVAD is able to reach an AUC score of 82.17\%, surpassing the current state-of-the-art by +2.87 AUC. Our code will be available on https://github.com/IMPLabUniPr/movad/tree/movad_vad
Anomaly detection in videos has been attracting an increasing amount of attention. Despite the competitive performance of recent methods on benchmark datasets, they typically lack desirable features such as modularity, cross-domain adaptivity, interpretability, and real-time anomalous event detection. Furthermore, current state-of-the-art approaches are evaluated using the standard instance-based detection metric by considering video frames as independent instances, which is not ideal for video anomaly detection. Motivated by these research gaps, we propose a modular and unified approach to the online video anomaly detection and localization problem, called MOVAD, which consists of a novel transfer learning based plug-and-play architecture, a sequential anomaly detector, a mathematical framework for selecting the detection threshold, and a suitable performance metric for real-time anomalous event detection in videos. Extensive performance evaluations on benchmark datasets show that the proposed framework significantly outperforms the current state-of-the-art approaches.