Research map / Vision-language and multimodal / Human and embodied modeling
Toward a Taxonomy and Computational Models of Abnormalities in Images
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| add the object-centric surprise score to the combined model (ablation) | Three-score abnormality model · Our abnormal image dataset | AUC: 0.8004 -> 0.8914 | Helped |
| add the scene-centric surprise score to the combined model (ablation) | Three-score abnormality model · Our abnormal image dataset | AUC: 0.8635 -> 0.8914 | Helped |
| add the context-centric surprise score to the combined model (ablation) | Three-score abnormality model · Our abnormal image dataset | AUC: 0.8863 -> 0.8914 | Helped |
| add attribute classifier accuracy term to object-centric score (ablation) | Object-centric surprise score · Object-centric AUC | AUC: 0.8050 -> 0.8662 | Helped |
| add attribute relevance term to object-centric score (ablation) | Object-centric surprise score · Object-centric AUC | AUC: 0.7985 -> 0.8662 | Helped |
| add object location surprise I(L|O) to co-occurrence in context score (ablation) | Context-centric surprise score · Context-centric AUC | AUC: 0.6778 -> 0.8255 | Helped |
| add relative size modulation Lambda(O) to context score (ablation) | Context-centric surprise score · Context-centric AUC | AUC: 0.6778 -> 0.6923 | Helped |
| add relevance and accuracy terms to scene-centric score (ablation) | Scene-centric surprise score · Scene-centric AUC | AUC: 0.6625 -> 0.7418 | Helped |
| add parametric inverse Gaussian model over raw surprise scores (ablation) | Abnormality reasoning · Our abnormal image dataset | KL-divergence (lower better): 2.276 -> 0.8142 | Helped |
| use proposed generative model trained on normal images vs two-class SVM on deep features trained with abnormal images | Abnormality classification · Our abnormal image dataset | AUC: 0.8524 -> 0.8914 | Helped |
| proposed object-centric model vs prior object-centric baseline | Object-centric abnormality · Exp I | AUC: 0.9125 -> 0.9311 | Helped |
| proposed context-centric model vs prior context baseline | Context-centric abnormality · Exp II | AUC: 0.8518 -> 0.8943 | Helped |
Closest papers by meaning
- Is this Generated Person Existed in Real-world? Fine-grained Detecting and Calibrating Abnormal Human-body (CVPR 2025 · similarity 0.85)
- UniVAD: A Training-free Unified Model for Few-shot Visual Anomaly Detection (CVPR 2025 · similarity 0.83)
- Odd-One-Out: Anomaly Detection by Comparing with Neighbors (CVPR 2025 · similarity 0.83)
- MANTA: A Large-Scale Multi-View and Visual-Text Anomaly Detection Dataset for Tiny Objects (CVPR 2025 · similarity 0.83)
- Towards Zero-Shot Anomaly Detection and Reasoning with Multimodal Large Language Models (CVPR 2025 · similarity 0.82)
- Adapting Visual-Language Models for Generalizable Anomaly Detection in Medical Images (CVPR 2024 · similarity 0.82)
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.