Research map / Label-efficient and robust learning / OOD and anomaly detection
ATTA: Anomaly-aware Test-Time Adaptation for Out-of-Distribution Detection in Segmentation
Zhitong Gao, Shipeng Yan, Xuming He
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add ATTA test-time adaptation | PEBAL · Road Anomaly | AUC: 87.63 -> 92.11 | Improved |
| add ATTA test-time adaptation | PEBAL · FS Static-C | AUC: 67.63 -> 99.21 | Improved |
| add ATTA to Max Logit | Max Logit · FS Static | AUC: 95.66 -> 95.48 | No clear effect |
| add ATTA to Max Logit | Max Logit · Road Anomaly | AUC: 72.78 -> 76.60 | Improved |
| add ATTA to Energy | Energy · Road Anomaly | AP: 19.54 -> 25.27 | Improved |
| add TBN or Tent instead of ATTA | PEBAL · FS Static | AUC: 99.61 -> 99.05 | Hurt |
| add Tent instead of ATTA | PEBAL · Road Anomaly | FPR95: 44.58 -> 60.0 | Hurt |
| add selective batch norm (SBN) (ablation) | PEBAL · Road Anomaly | AUC: 87.63 -> 90.84 | Improved |
| add anomaly-aware self-training (AST) (ablation) | PEBAL · Road Anomaly | AUC: 87.63 -> 88.72 | Improved |
| use both batch and training statistics in SBN rather than batch only (ablation) | ATTA · Road Anomaly | AUC: 86.29 -> 92.11 | Improved |
| anomaly-aware entropy instead of seen-class-only entropy (ablation) | ATTA · Road Anomaly | AUC: 90.46 -> 92.11 | Improved |
| GMM-based normalization instead of z-score (ablation) | ATTA · Road Anomaly | AUC: 91.25 -> 92.11 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Tent — Tent: Fully Test-Time Adaptation by Entropy Minimization (ICLR 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Rethinking the Evaluation of Out-of-Distribution Detection: A Sorites Paradox (NeurIPS 2024 · similarity 0.90)
- ImageNet-OOD: Deciphering Modern Out-of-Distribution Detection Algorithms (ICLR 2024 · similarity 0.90)
- Extremely Simple Multimodal Outlier Synthesis for Out-of-Distribution Detection and Segmentation (NeurIPS 2025 · similarity 0.89)
- Generalize or Detect? Towards Robust Semantic Segmentation Under Multiple Distribution Shifts (NeurIPS 2024 · similarity 0.89)
- A Systematic Analysis of Out-of-Distribution Detection Under Representation and Training Paradigm Shifts (NeurIPS 2026 · similarity 0.88)
- Image Background Serves as Good Proxy for Out-of-distribution Data (ICLR 2024 · similarity 0.87)
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.
