Video Surveillance
Overview
Video surveillance research builds systems that detect people, vehicles, and events, track activity over time, and help operators search or review camera footage. Common approaches include object detectors and trackers, pose and motion analysis, weakly supervised anomaly detection, and newer vision-language models for describing scenes and answering questions. Methods are tested on settings such as traffic monitoring and anomaly benchmarks including UCF-Crime, XD-Violence, and ShanghaiTech, with growing attention to real-time edge deployment and privacy-preserving processing. A central challenge is making results reliable beyond familiar datasets: camera angle, image quality, location, and rare events can undermine performance, while privacy, useful detail, and human oversight must be balanced.
Latest papers
66 papers3 papers in the last four weeks, down 57% on the four weeks before. 0.0% of all new papers.
From Weapons to Events: Evaluating Evidential Support in Multimodal AI for Policing
CVBL-CCTV-Face Dataset for Real-World Face Enhancement and Recognition in Surveillance
Auditing Observation-Quality Shortcuts in the Evaluation of Person-Centered Motion for Video Anomaly Detection
Video-Based Sexual Harassment Detection Using Fine-Tuned Video Transformers
VESSI - VLM-Enhanced Support for Surveillance and Investigations
CCDF: A Benchmark Dataset for Deepfake Detection in Surveillance Footage
Can We Anticipate Violence? Multimodal Learning from Pre-Incident Behavioral Cues
Decision Readouts for Text-Mediated Video Anomaly Detection: An Exploratory Evaluation of Jev and Qwen
Copy What Is Seen, Generate What Is Not: Training-Free Anomaly-Aware Video Restoration
CatchMonitor: a machine learning system for automated fish discard quantification
BodyCam-VQA: Enhanced Body-Worn Camera Video Captioning via Multimodal Reasoning and Probe Question Generation
Data-Efficient Crosswalk Segmentation from Overhead CCTV via Confidence- and Geometry-Guided Pseudo-Labeling
Parser-Free VLM Verification for Federated Weakly Supervised Video Anomaly Detection
Federated Binary Gating with Server-Side Vision-Language Inference for Surveillance Anomaly Classification
Thinking with Cameras: Active Visual Reasoning via Dynamic Viewpoint Control for Surveillance Video Understanding
A Benchmark for Vehicle Attribute Classification in Cross-Domain Surveillance Scenarios
STAIN-FL: Stealthy Targeted Attack Injection with Contextual Triggers in Federated Learning
UCF-Crime using 1024-dimensional I3D features in a non-IID four-client multi-agency setting, comparing FedAvg and FedProx under sparse and continuous attacks. Results show that sparse attacks have low-detectability, operationally significant attacks rather than high-intensity attacks: they keep the mean clean-accuracy drop below , yet still misclassify more than half of triggered anomalies at peak backdoor accuracy under FedAvg () and FedProx (). Under FedAvg, the sparse backdoor remains above the backdoor-accuracy threshold for an average of post-attack rounds, highlighting the persistence risk of contextually triggered attacks in surveillance systems.