Environment-Invariant Curriculum Relation Learning for Fine-Grained Scene Graph Generation

TL;DR

This paper introduces EICR, a novel method for scene graph generation that addresses both predicate and context imbalance issues by learning environment-invariant relations and applying curriculum learning, leading to improved results.

Contribution

The paper proposes a plug-and-play EICR framework that tackles subject-object imbalance and predicate imbalance in SGG through environment-invariant learning and curriculum strategies.

Findings

Significant performance improvements on VG and GQA datasets.

EICR is compatible with various existing SGG models.

Effectively addresses class and context imbalance issues.

Abstract

The scene graph generation (SGG) task is designed to identify the predicates based on the subject-object pairs.However,existing datasets generally include two imbalance cases: one is the class imbalance from the predicted predicates and another is the context imbalance from the given subject-object pairs, which presents significant challenges for SGG. Most existing methods focus on the imbalance of the predicted predicate while ignoring the imbalance of the subject-object pairs, which could not achieve satisfactory results. To address the two imbalance cases, we propose a novel Environment Invariant Curriculum Relation learning (EICR) method, which can be applied in a plug-and-play fashion to existing SGG methods. Concretely, to remove the imbalance of the subject-object pairs, we first construct different distribution environments for the subject-object pairs and learn a model invariant to the environment changes. Then, we construct a class-balanced curriculum learning strategy to balance the different environments to remove the predicate imbalance. Comprehensive experiments conducted on VG and GQA datasets demonstrate that our EICR framework can be taken as a general strategy for various SGG models, and achieve significant improvements.