AI Paper Atlas

Research map / Video / Video segmentation and tracking

HeadHunt-VAD: Hunting Robust Anomaly-Sensitive Heads in MLLM for Tuning-Free Video Anomaly Detection

Zhaolin Cai, Fan Li, Ziwei Zheng, Haixia Bi, Lijun He

AAAI 2026 · Video › Video segmentation and tracking · approach: Weakly supervised video anomaly detection · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
probe attention-head features instead of aggregated final-layer features (ablation)HeadHunt-VAD · UCF-CrimeAUC (%): 72.10 -> 87.03Improved
select heads with Robust Head Identification instead of random K heads (ablation)HeadHunt-VAD · UCF-CrimeAUC (%): 66.65 -> 87.03Improved
use RHI across five prompts instead of a single coarse prompt (ablation)HeadHunt-VAD · UCF-CrimeAUC (%): 81.86 -> 87.03Improved
use manually crafted oracle prompt instead of RHI (ablation)HeadHunt-VAD · UCF-CrimeAUC (%): 87.03 -> 87.11No clear effect
replace logistic regression scorer with 2-layer MLP (ablation)HeadHunt-VAD · UCF-Crime / XD-ViolenceAUC / AP: +0.22% AUC, +0.18% APNo clear effect
replace logistic regression scorer with linear SVM (ablation)HeadHunt-VAD · UCF-CrimeAUC (%): 87.03 -> 84.95Hurt
add Gaussian smoothing in temporal locator (ablation)HeadHunt-VAD · UCF-CrimeAUC (%): 82.44 -> 87.03Improved
calibrate threshold on validation set instead of fixed values (ablation)HeadHunt-VAD · UCF-CrimeAUC (%): 80.32 -> 87.03 (best fixed tau=0.50)Improved
use head-level probing instead of text generation (LAVAD, LLaVA-1.5)HeadHunt-VAD · UCF-CrimeAUC (%): 80.28 -> 87.03 (LAVAD)Improved
single forward pass over top-K heads instead of autoregressive decodingHeadHunt-VAD · inference efficiencyspeed: Speed:+89.8%Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy