Steering and Rectifying Latent Representation Manifolds in Frozen Multi-modal LLMs for Video Anomaly Detection

• URL: http://arxiv.org/abs/2602.24021v1
• Date: Fri, 27 Feb 2026 13:48:50 GMT
• Title: Steering and Rectifying Latent Representation Manifolds in Frozen Multi-modal LLMs for Video Anomaly Detection
• Authors: Zhaolin Cai, Fan Li, Huiyu Duan, Lijun He, Guangtao Zhai,
• Abstract summary: Video anomaly detection (VAD) aims to identify abnormal events in videos.<n>We propose SteerVAD, which advances MLLM-based VAD by shifting from passively reading to actively steering and rectifying internal representations.<n>Our method achieves state-of-the-art performance among tuning-free approaches requiring only 1% of training data.
• Score: 52.5174167737992
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Video anomaly detection (VAD) aims to identify abnormal events in videos. Traditional VAD methods generally suffer from the high costs of labeled data and full training, thus some recent works have explored leveraging frozen multi-modal large language models (MLLMs) in a tuning-free manner to perform VAD. However, their performance is limited as they directly inherit pre-training biases and cannot adapt internal representations to specific video contexts, leading to difficulties in handling subtle or ambiguous anomalies. To address these limitations, we propose a novel intervention framework, termed SteerVAD, which advances MLLM-based VAD by shifting from passively reading to actively steering and rectifying internal representations. Our approach first leverages the gradient-free representational separability analysis (RSA) to identify top attention heads as latent anomaly experts (LAEs) which are most discriminative for VAD. Then a hierarchical meta-controller (HMC) generates dynamic rectification signals by jointly conditioning on global context and these LAE outputs. The signals execute targeted, anisotropic scaling directly upon the LAE representation manifolds, amplifying anomaly-relevant dimensions while suppressing inherent biases. Extensive experiments on mainstream benchmarks demonstrate our method achieves state-of-the-art performance among tuning-free approaches requiring only 1% of training data, establishing it as a powerful new direction for video anomaly detection. The code will be released upon the publication.

Related papers

• HeadHunt-VAD: Hunting Robust Anomaly-Sensitive Heads in MLLM for Tuning-Free Video Anomaly Detection [9.217348688177298]
  Video Anomaly Detection (VAD) aims to locate events that deviate from normal patterns in videos.<n>Recent tuning-free methods based on Multimodal Large Language Models (MLLMs) offer a promising alternative by leveraging their rich world knowledge.<n>We propose HeadHunt-VAD, a novel tuning-free VAD paradigm that bypasses textual generation by directly hunting robust anomaly-sensitive internal attention heads.
  arXiv  Detail & Related papers  (2025-12-19T14:07:34Z)
• Steering Vision-Language-Action Models as Anti-Exploration: A Test-Time Scaling Approach [78.4812458793128]
  We propose textbfTACO, a test-time-scaling framework that applies a lightweight pseudo-count estimator as a high-fidelity verifier of action chunks.<n>Our method resembles the classical anti-exploration principle in offline reinforcement learning (RL), and being gradient-free, it incurs significant computational benefits.
  arXiv  Detail & Related papers  (2025-12-02T14:42:54Z)
• Contamination Detection for VLMs using Multi-Modal Semantic Perturbation [73.76465227729818]
  Open-source Vision-Language Models (VLMs) have achieved state-of-the-art performance on benchmark tasks.<n>Pretraining corpora raise a critical concern for both practitioners and users: inflated performance due to test-set leakage.<n>We show that existing detection approaches either fail outright or exhibit inconsistent behavior.<n>We propose a novel simple yet effective detection method based on multi-modal semantic perturbation.
  arXiv  Detail & Related papers  (2025-11-05T18:59:52Z)
• Unified Unsupervised Anomaly Detection via Matching Cost Filtering [113.43366521994396]
  Unsupervised anomaly detection (UAD) aims to identify image- and pixel-level anomalies using only normal training data.<n>We present Unified Cost Filtering (UCF), a generic post-hoc refinement framework for refining anomaly cost volume of any UAD model.
  arXiv  Detail & Related papers  (2025-10-03T03:28:18Z)
• HiProbe-VAD: Video Anomaly Detection via Hidden States Probing in Tuning-Free Multimodal LLMs [8.18063726177317]
  Video Anomaly Detection (VAD) aims to identify and locate deviations from normal patterns in video sequences.<n>We propose HiProbe-VAD, a novel framework that leverages pre-trained Multimodal Large Language Models (MLLMs) for VAD without requiring fine-tuning.
  arXiv  Detail & Related papers  (2025-07-23T10:41:46Z)
• Beyond Fully Supervised Pixel Annotations: Scribble-Driven Weakly-Supervised Framework for Image Manipulation Localization [11.10178274806454]
  We propose a form of weak supervision that improves the annotation efficiency and detection performance.<n>We re-annotated mainstream IML datasets with scribble labels and propose the first scribble-based IML dataset.<n>We employ self-supervised training with a structural consistency loss to encourage the model to produce consistent predictions.
  arXiv  Detail & Related papers  (2025-07-17T11:45:27Z)
• Multimodal LLM-Guided Semantic Correction in Text-to-Image Diffusion [52.315729095824906]
  MLLM Semantic-Corrected Ping-Pong-Ahead Diffusion (PPAD) is a novel framework that introduces a Multimodal Large Language Model (MLLM) as a semantic observer during inference.<n>It performs real-time analysis on intermediate generations, identifies latent semantic inconsistencies, and translates feedback into controllable signals that actively guide the remaining denoising steps.<n>Extensive experiments demonstrate PPAD's significant improvements.
  arXiv  Detail & Related papers  (2025-05-26T14:42:35Z)
• Vad-R1: Towards Video Anomaly Reasoning via Perception-to-Cognition Chain-of-Thought [58.321044666612174]
  Vad-R1 is an end-to-end MLLM-based framework for Video Anomaly Reasoning.<n>We design a Perception-to-Cognition Chain-of-Thought (P2C-CoT) that simulates the human process of recognizing anomalies.<n>We also propose an improved reinforcement learning algorithm AVA-GRPO, which explicitly incentivizes the anomaly reasoning capability of MLLMs.
  arXiv  Detail & Related papers  (2025-05-26T12:05:16Z)
• Appearance Blur-driven AutoEncoder and Motion-guided Memory Module for Video Anomaly Detection [14.315287192621662]
  Video anomaly detection (VAD) often learns the distribution of normal samples and detects the anomaly through measuring significant deviations. Most VADs cannot cope with cross-dataset validation for new target domains. We propose a novel VAD method with a motion-guided memory module to achieve cross-dataset validation with zero-shot.
  arXiv  Detail & Related papers  (2024-09-26T07:48:20Z)
• Unsupervised Video Anomaly Detection with Diffusion Models Conditioned on Compact Motion Representations [17.816344808780965]
  unsupervised video anomaly detection (VAD) problem involves classifying each frame in a video as normal or abnormal, without any access to labels. To accomplish this, proposed method employs conditional diffusion models, where the input data is features extracted from pre-trained network. Our method utilizes a data-driven threshold and considers a high reconstruction error as an indicator of anomalous events.
  arXiv  Detail & Related papers  (2023-07-04T07:36:48Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.