SOTAVerified

Unbiased Scene Graph Generation

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")

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 1–25 of 30 papers

TitleStatusHype
Unbiased Scene Graph Generation by Type-Aware Message Passing on Heterogeneous and Dual Graphs—0
Ensemble Predicate Decoding for Unbiased Scene Graph Generation—0
Semantic Diversity-aware Prototype-based Learning for Unbiased Scene Graph GenerationCode1
ALF: Adaptive Label Finetuning for Scene Graph Generation—0
FloCoDe: Unbiased Dynamic Scene Graph Generation with Temporal Consistency and Correlation Debiasing—0
Head-Tail Cooperative Learning Network for Unbiased Scene Graph GenerationCode0
Vision Relation Transformer for Unbiased Scene Graph GenerationCode1
Compositional Feature Augmentation for Unbiased Scene Graph GenerationCode1
Generalized Unbiased Scene Graph Generation—0
Unbiased Scene Graph Generation via Two-stage Causal Modeling—0
Multi-Label Meta Weighting for Long-Tailed Dynamic Scene Graph GenerationCode0
Unbiased Scene Graph Generation in VideosCode1
Skew Class-balanced Re-weighting for Unbiased Scene Graph GenerationCode0
Peer Learning for Unbiased Scene Graph Generation—0
Unbiased Scene Graph Generation using Predicate Similarities—0
Context-aware Mixture-of-Experts for Unbiased Scene Graph Generation—0
Label Semantic Knowledge Distillation for Unbiased Scene Graph Generation—0
Rethinking the Evaluation of Unbiased Scene Graph Generation—0
Dual-branch Hybrid Learning Network for Unbiased Scene Graph GenerationCode1
Learning Structured Representations of Visual Scenes—0
Fine-Grained Scene Graph Generation with Data TransferCode1
Stacked Hybrid-Attention and Group Collaborative Learning for Unbiased Scene Graph GenerationCode1
Resistance Training using Prior Bias: toward Unbiased Scene Graph GenerationCode1
PPDL: Predicate Probability Distribution Based Loss for Unbiased Scene Graph Generation—0
Recovering the Unbiased Scene Graphs from the Biased OnesCode1
Show:102550
← PrevPage 1 of 2Next →

No leaderboard results yet.