# Evaluation of Vision-LLMs in Surveillance Video

arXiv: https://arxiv.org/abs/2510.23190 · HTML: https://mlsanity.com/paper/2510.23190/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 14 of its 23 references that are in the map.
- First posted on arXiv 27 Oct 2025.

## Builds on (most cited first, 14 of 14)

- [BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension](https://mlsanity.com/paper/1910.13461/) (2019)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [Real-world Anomaly Detection in Surveillance Videos](https://mlsanity.com/paper/1801.04264/) (2018)
- [VideoLLaMA 3: Frontier Multimodal Foundation Models for Image and Video Understanding](https://mlsanity.com/paper/2501.13106/) (2025)
- [Not only Look, but also Listen: Learning Multimodal Violence Detection under Weak Supervision](https://mlsanity.com/paper/2007.04687/) (2020)
- [Harnessing Large Language Models for Training-free Video Anomaly Detection](https://mlsanity.com/paper/2404.01014/) (2024)
- [Holmes-VAD: Towards Unbiased and Explainable Video Anomaly Detection via Multi-modal LLM](https://mlsanity.com/paper/2406.12235/) (2024)
- [Open-Vocabulary Video Anomaly Detection](https://mlsanity.com/paper/2311.07042/) (2023)
- [DeepPrivacy2: Towards Realistic Full-Body Anonymization](https://mlsanity.com/paper/2211.09454/) (2022)
- [RWF-2000: An Open Large Scale Video Database for Violence Detection](https://mlsanity.com/paper/1911.05913/) (2019)
- [Delving into CLIP latent space for Video Anomaly Recognition](https://mlsanity.com/paper/2310.02835/) (2023)
- [MissionGNN: Hierarchical Multimodal GNN-based Weakly Supervised Video Anomaly Recognition with Mission-Specific Knowledge Graph Generation](https://mlsanity.com/paper/2406.18815/) (2024)
- [Masked Generative Priors Improve World Models Sequence Modelling Capabilities](https://mlsanity.com/paper/2410.07836/) (2024)
- [Object-Centric Temporal Consistency via Conditional Autoregressive Inductive Biases](https://mlsanity.com/paper/2410.15728/) (2024)

## Built on by (most cited first, 3 of 3)

- [Physical Prompt Injection Attacks on Large Vision-Language Models](https://mlsanity.com/paper/2601.17383/) (2026)
- [Vision-Language Models Suppress Female Representations Under Ambiguous Input](https://mlsanity.com/paper/2605.31556/) (2026)
- [BodyCam-VQA: Enhanced Body-Worn Camera Video Captioning via Multimodal Reasoning and Probe Question Generation](https://mlsanity.com/paper/2609.10815/) (2026)

## Abstract (arXiv)

The widespread use of cameras in our society has created an overwhelming amount of video data, far exceeding the capacity for human monitoring. This presents a critical challenge for public safety and security, as the timely detection of anomalous or criminal events is crucial for effective response and prevention. The ability for an embodied agent to recognize unexpected events is fundamentally tied to its capacity for spatial reasoning. This paper investigates the spatial reasoning of vision-language models (VLMs) by framing anomalous action recognition as a zero-shot, language-grounded task, addressing the embodied perception challenge of interpreting dynamic 3D scenes from sparse 2D video. Specifically, we investigate whether small, pre-trained vision--LLMs can act as spatially-grounded, zero-shot anomaly detectors by converting video into text descriptions and scoring labels via textual entailment. We evaluate four open models on UCF-Crime and RWF-2000 under prompting and privacy-preserving conditions. Few-shot exemplars can improve accuracy for some models, but may increase false positives, and privacy filters -- especially full-body GAN transforms -- introduce inconsistencies that degrade accuracy. These results chart where current vision--LLMs succeed (simple, spatially salient events) and where they falter (noisy spatial cues, identity obfuscation). Looking forward, we outline concrete paths to strengthen spatial grounding without task-specific training: structure-aware prompts, lightweight spatial memory across clips, scene-graph or 3D-pose priors during description, and privacy methods that preserve action-relevant geometry. This positions zero-shot, language-grounded pipelines as adaptable building blocks for embodied, real-world video understanding. Our implementation for evaluating VLMs is publicly available at: https://github.com/pascalbenschopTU/VLLM_AnomalyRecognition