Papers with Code · CC BY-SA 4.0

Unbiased Scene Graph Generation

Scene Parsing › Scene Graph Generation › Unbiased Scene Graph Generation 31 평가 기록 1 datasets

Unbiased Scene Graph Generation (Unbiased SGG) aims to predict more informative scene graphs composed of more "tail predicates" *(in contrast to "head predicates" in terms of class frequencies) by dealing with the skewed, long-tailed predicate class distribution. (Definition from Chiou et al. "Recovering the Unbiased Scene Graphs from the Biased Ones")

논문

PCPL: Predicate-Correlation Perception Learning for Unbiased Scene Graph Generation

2009.00893 2020-09-02 이 페이지의 평가 6건

ng-mR@20: 25.6mR@20: 19.3F@100: 35.7

Recovering the Unbiased Scene Graphs from the Biased Ones

2107.02112 2021-07-05 이 페이지의 평가 6건

ng-mR@20: 30.0mR@20: 22.1F@100: 37.6

Unbiased Scene Graph Generation from Biased Training

2002.11949 2020-02-27 이 페이지의 평가 6건

ng-mR@20: 20.9mR@20: 19.2F@100: 36.9

CogTree: Cognition Tree Loss for Unbiased Scene Graph Generation

2009.07526 2020-09-16 이 페이지의 평가 5건

mR@20: 22.0F@100: 35.9

Fine-Grained Scene Graph Generation with Data Transfer

2203.11654 2022-03-22 이 페이지의 평가 3건

ng-mR@20: 36.0mR@20: 28.9F@100: 44.1

Semantic Diversity-aware Prototype-based Learning for Unbiased Scene Graph Generation

2407.15396 2024-07-22 이 페이지의 평가 3건

ng-mR@20: 31.3mR@20: 26.2F@100: 44.9

Stacked Hybrid-Attention and Group Collaborative Learning for Unbiased Scene Graph Generation

2203.09811 2022-03-18 이 페이지의 평가 1건

mR@20: 35.6