Bipartite Graph Network with Adaptive Message Passing for Unbiased Scene Graph Generation (2021-04-01T00:00:00.000000Z)

TL;DR

This work introduces a novel confidence-aware bipartite graph neural network with adaptive message propagation mechanism for unbiased scene graph generation and proposes an efficient bi-level data resampling strategy to alleviate the imbalanced data distribution problem in training the authors' graph network.

Abstract

Scene graph generation is an important visual understanding task with a broad range of vision applications. Despite recent tremendous progress, it remains challenging due to the intrinsic long-tailed class distribution and large intra-class variation. To address these issues, we introduce a novel confidence-aware bipartite graph neural network with adaptive message propagation mechanism for unbiased scene graph generation. In addition, we propose an efficient bi-level data resampling strategy to alleviate the imbalanced data distribution problem in training our graph network. Our approach achieves superior or competitive performance over previous methods on several challenging datasets, including Visual Genome, Open Images V4/V6, demonstrating its effectiveness and generality.

Authors

Songyang Zhang

4 papers

Rongjie Li

1 papers

Bo Wan

2 papers

TL;DR

Abstract

Authors

References63 items

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

Tackling the Unannotated: Scene Graph Generation with Bias-Reduced Models

Sketching Image Gist: Human-Mimetic Hierarchical Scene Graph Generation

Scene Graph Reasoning for Visual Question Answering

Overcoming Classifier Imbalance for Long-Tail Object Detection With Balanced Group Softmax

Learning Visual Commonsense for Robust Scene Graph Generation

Graph Density-Aware Losses for Novel Compositions in Scene Graph Generation

Measuring and Improving the Use of Graph Information in Graph Neural Networks

Learning to Segment the Tail

GPS-Net: Graph Property Sensing Network for Scene Graph Generation

Equalization Loss for Long-Tailed Object Recognition

Unbiased Scene Graph Generation From Biased Training

NODIS: Neural Ordinary Differential Scene Understanding

Bridging Knowledge Graphs to Generate Scene Graphs

BBN: Bilateral-Branch Network With Cumulative Learning for Long-Tailed Visual Recognition

Decoupling Representation and Classifier for Long-Tailed Recognition

Dynamically Pruned Message Passing Networks for Large-Scale Knowledge Graph Reasoning

Learning Imbalanced Datasets with Label-Distribution-Aware Margin Loss

LVIS: A Dataset for Large Vocabulary Instance Segmentation

Exploring Context and Visual Pattern of Relationship for Scene Graph Generation

LatentGNN: Learning Efficient Non-local Relations for Visual Recognition

Large-Scale Long-Tailed Recognition in an Open World

Scene Graph Generation With External Knowledge and Image Reconstruction

Heterogeneous Graph Attention Network

Knowledge-Embedded Routing Network for Scene Graph Generation

Graphical Contrastive Losses for Scene Graph Generation

Back Propagation

Class-Balanced Loss Based on Effective Number of Samples

Detecting Unseen Visual Relations Using Analogies

Auto-Encoding Scene Graphs for Image Captioning

Explainable and Explicit Visual Reasoning Over Scene Graphs

Learning to Compose Dynamic Tree Structures for Visual Contexts

Grid R-CNN

Attentive Relational Networks for Mapping Images to Scene Graphs

LinkNet: Relational Embedding for Scene Graph

Graph R-CNN for Scene Graph Generation

Zoom-Net: Mining Deep Feature Interactions for Visual Relationship Recognition

Factorizable Net: An Efficient Subgraph-based Framework for Scene Graph Generation

Rethinking Knowledge Graph Propagation for Zero-Shot Learning

Exploring the Limits of Weakly Supervised Pretraining

Dynamic Few-Shot Visual Learning Without Forgetting

Deep Collective Classification in Heterogeneous Information Networks

Non-local Neural Networks

Neural Motifs: Scene Graph Parsing with Global Context

Scene Graph Generation from Objects, Phrases and Region Captions

Attention is All you Need

Inductive Representation Learning on Large Graphs

Detecting Visual Relationships with Deep Relational Networks

Scene Graph Generation by Iterative Message Passing

Aggregated Residual Transformations for Deep Neural Networks

Graph-Structured Representations for Visual Question Answering

Semi-Supervised Classification with Graph Convolutional Networks

Visual Genome: Connecting Language and Vision Using Crowdsourced Dense Image Annotations

SSD: Single Shot MultiBox Detector

Gated Graph Sequence Neural Networks

Cost-Sensitive Learning of Deep Feature Representations From Imbalanced Data

Image retrieval using scene graphs

Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

The open images dataset v4: Unified image classification, object detection, and visual relationship detection at scale

The Graph Neural Network Model

C4.5, Class Imbalance, and Cost Sensitivity: Why Under-Sampling beats Over-Sampling

SMOTE: Synthetic Minority Over-sampling Technique

Overcoming classiﬁer im-9

Field of Study

Journal Information

Name

Page

Venue Information

Name

Type

URL

Alternate Names