mlsanityData to 14 Oct 2026

Unbiased Video Scene Graph Generation via Visual and Semantic Dual Debiasing

arXiv 2503.00548 ↗First posted 1 Mar 2025cs.CV cs.MMcited by 6

Not on the map. Why?

Abstract (arXiv)

Video Scene Graph Generation (VidSGG) aims to capture dynamic relationships among entities by sequentially analyzing video frames and integrating visual and semantic information. However, VidSGG is challenged by significant biases that skew predictions. To mitigate these biases, we propose a VIsual and Semantic Awareness (VISA) framework for unbiased VidSGG. VISA addresses visual bias through memory-enhanced temporal integration that enhances object representations and concurrently reduces semantic bias by iteratively integrating object features with comprehensive semantic information derived from triplet relationships. This visual-semantics dual debiasing approach results in more unbiased representations of complex scene dynamics. Extensive experiments demonstrate the effectiveness of our method, where VISA outperforms existing unbiased VidSGG approaches by a substantial margin (e.g., +13.1% improvement in mR@20 and mR@50 for the SGCLS task under Semi Constraint).

Key facts. Cited by 6 papers in the map (data through 14 Oct 2026). Builds on 19 of its 23 references that are in the map. First posted on arXiv 1 Mar 2025.

Builds on

19 of 23 references.

References · 19

  1. 2017Attention Is All You Needcited by 60,361
  2. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  3. 2015Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networkscited by 9,639
  4. 2017Neural Motifs: Scene Graph Parsing with Global Contextcited by 414
  5. 2020Unbiased Scene Graph Generation from Biased Trainingcited by 356
  6. 2018Learning to Compose Dynamic Tree Structures for Visual Contextscited by 259
  7. 2019Action Genome: Actions as Composition of Spatio-temporal Scene Graphscited by 250
  8. 2021SGTR: End-to-end Scene Graph Generation with Transformercited by 103
  9. 2022RelTR: Relation Transformer for Scene Graph Generationcited by 96
  10. 2020PCPL: Predicate-Correlation Perception Learning for Unbiased Scene Graph Generationcited by 75
  11. 2022Relationformer: A Unified Framework for Image-to-Graph Generationcited by 67
  12. 2015Seeing through the Human Reporting Bias: Visual Classifiers from Noisy Human-Centric Labelscited by 64
7 more
  1. 2021Target Adaptive Context Aggregation for Video Scene Graph Generationcited by 52
  2. 2023Panoptic Video Scene Graph Generationcited by 46
  3. 2023Unbiased Scene Graph Generation in Videoscited by 39
  4. 20244D Panoptic Scene Graph Generationcited by 22
  5. 2021Social Fabric: Tubelet Compositions for Video Relation Detectioncited by 18
  6. 2022Label Semantic Knowledge Distillation for Unbiased Scene Graph Generationcited by 14
  7. 2022Dual-branch Hybrid Learning Network for Unbiased Scene Graph Generationcited by 10

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 4; 2026: 2.

6 papers.

  1. 2026Synthetic Visual Genome 2: Extracting Large-scale Spatio-Temporal Scene Graphs from Videoscited by 3
  2. 2025UNO: Unifying One-stage Video Scene Graph Generation via Object-Centric Visual Representation Learningcited by 2
  3. 2025Language-Driven Object-Oriented Two-Stage Method for Scene Graph Anticipationcited by 1
  4. 2025A Closer Look at Dynamic Scene Graph Generation In the Era of Multimodal Large Language Modelscited by 0
  5. 2025Click2Graph: Interactive Panoptic Video Scene Graphs from a Single Clickcited by 0
  6. 2026T-STAR: A Large-Scale Benchmark for Spatio-Temporal Panoptic Scene Graph Generation in Satellite Videocited by 0