AI Paper Atlas

Research map / Benchmarks / Road Anomaly

Road Anomaly leaderboard

14 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on Road Anomaly. Best AP: 98.66 (Ours+DaCUP (recon.), ECCV 2024).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

Road Anomaly · AP

14 papers, each with its best reported number for its own method; higher is better.

#AP ↑MethodPaperTopic
198.66Ours+DaCUP (recon.)
DINOv2 ViT-L
PixOOD: Pixel-Level Out-of-Distribution Detection
ECCV 2024 · Table 2
OOD and anomaly detection
294.4JSR-Net (checkp2)
DeepLabV3 ResNet-101
Road Anomaly Detection by Partial Image Reconstruction With Segmentation Coupling
ICCV 2021 · Figure 3
Open-set and few-shot detection
392.86UEM (G-D)
DINOv2-b
A Likelihood Ratio-Based Approach to Segmenting Unknown Objects
IJCV 2025 · Table 3
OOD and anomaly detection
490.47RbA + ClimaOoD
RbA
ClimaOoD: Improving Anomaly Segmentation via Physically Realistic Synthetic Data
CVPR 2026 · Table 2
Open-set and few-shot detection
590.17Ours
Mask2Former
Generalize or Detect? Towards Robust Semantic Segmentation Under Multiple Distribution Shifts
NeurIPS 2024 · Table 1
Open-set and few-shot detection
689.1DOoD (Segmenter-B)
Segmenter-B
Diffusion for Out-of-Distribution Detection on Road Scenes and Beyond
ECCV 2024 · Table (state-of-the-art compar
OOD and anomaly detection
788.41Mask2Former + Prompt2Seg
Mask2Former
Prompting Diffusion Models for Zero-Shot Instance Segmentation
NeurIPS 2026 · Table 2
Open-set and few-shot detection
887.34RbA + RWPM
RbA
Random Walk on Pixel Manifolds for Anomaly Segmentation of Complex Driving Scenes
ECCV 2024 · Table 1
Open-set and few-shot detection
962.37PEBAL (Ours)
ResNet101
Pixel-Wise Energy-Biased Abstention Learning for Anomaly Segmentation on Complex Urban Driving Scenes
ECCV 2022 · Table 4
Open-set and few-shot detection
1061.38ObsNet v2 + CSL₂
ObsNet v2
CSL: Class-Agnostic Structure-Constrained Learning for Segmentation Including the Unseen
AAAI 2024 · Table: SMIYC (Anomaly Track) a
Open-set and few-shot detection
4 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.