Video Anomaly Detection Through Spatial–Temporal Feature Relocalization and Calibrated Trajectory Modeling
Open Access
- 13 March 2026
- journal article
- research article
- Published by MDPI AG in Electronics
- Vol. 15 (6) , 1199
Abstract
To address the limitations of existing video anomaly detection methods that overly rely on pixel-space reconstruction and are sensitive to background noise and object scale variations, a self-supervised contrastive learning approach that integrates spatial–temporal feature relocalization with camera-calibrated trajectory modeling is proposed. The proposed method takes spatial–temporal feature relocalization as the core task and constructs a feature-level contrastive learning mechanism to guide the model to focus on discriminative local appearance variations and global temporal semantic evolution. While suppressing background interference and scale-related noise, the method enhances the modeling of fine-grained appearance anomalies and global action-related temporal anomalies. Furthermore, camera calibration is introduced to recover continuous object trajectories in physical space, and a temporal aggregation module is designed to jointly model object motion patterns in pixel space and physical space, thereby improving the model’s ability to perceive complex anomalous behaviors. Experimental results on multiple public video anomaly detection benchmarks demonstrate that the proposed method consistently outperforms existing approaches, validating its effectiveness and generalization capability.Keywords
This publication has 47 references indexed in Scilit:
- Unveiling Context-Related Anomalies: Knowledge Graph Empowered Decoupling of Scene and Action for Human-Related Video Anomaly DetectionIEEE Transactions on Circuits and Systems for Video Technology, 2025
- Dual-Stream Anomaly Detection Network for Real-World Traffic ScenariosIEEE Transactions on Intelligent Transportation Systems, 2024
- VADiffusion: Compressed Domain Information Guided Conditional Diffusion for Video Anomaly DetectionIEEE Transactions on Circuits and Systems for Video Technology, 2024
- RoFormer: Enhanced transformer with Rotary Position EmbeddingNeurocomputing, 2023
- Multi-Branch GAN-Based Abnormal Events Detection via Context Learning in Surveillance VideosIEEE Transactions on Circuits and Systems for Video Technology, 2023
- TransCNN: Hybrid CNN and transformer mechanism for surveillance anomaly detectionEngineering Applications of Artificial Intelligence, 2023
- PseudoBound: Limiting the anomaly reconstruction capability of one-class classifiers using pseudo anomaliesNeurocomputing, 2023
- Bidirectional Spatio-Temporal Feature Learning With Multiscale Evaluation for Video Anomaly DetectionIEEE Transactions on Circuits and Systems for Video Technology, 2022
- Fully Automatic Roadside Camera Calibration for Traffic SurveillanceIEEE Transactions on Intelligent Transportation Systems, 2014
- A flexible new technique for camera calibrationIEEE Transactions on Pattern Analysis and Machine Intelligence, 2000