arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2211.04773cs.CV
收藏

SG-Shuffle:用于场景图生成的多方位Shuffle Transformer

SG-Shuffle: Multi-aspect Shuffle Transformer for Scene Graph Generation

The University of Sydney(悉尼大学)

Anh Duc Bui, Soyeon Caren Han, Josiah Poon

摘要:

场景图生成(SGG)为人类理解以及视觉理解任务提供了图像的全面表示。由于在可用的注释数据中的对象和谓词标签的长尾偏置问题,从当前方法生成的场景图可能偏向于常见的、非信息性的关系标签。关系有时可能是非互斥的,可以从几何关系或语义关系等多个角度进行描述,这使得预测最合适的关系标签更具挑战性。在这项工作中,我们提出了用于场景图生成的SG-Shuffle流水线,它有3个组件:1)并行Transformer编码器,它通过将关系标签分组到具有相似目的的组中来学习以更排他性的方式预测对象关系; 2)Shuffle Transformer,它学习从上一步生成的类别特定特征中选择最终的关系标签; 3)加权的CE损失,用于减轻由不平衡数据集引起的训练偏差。

英文摘要:

Scene Graph Generation (SGG) serves a comprehensive representation of the images for human understanding as well as visual understanding tasks. Due to the long tail bias problem of the object and predicate labels in the available annotated data, the scene graph generated from current methodologies can be biased toward common, non-informative relationship labels. Relationship can sometimes be non-mutually exclusive, which can be described from multiple perspectives like geometrical relationships or semantic relationships, making it even more challenging to predict the most suitable relationship label. In this work, we proposed the SG-Shuffle pipeline for scene graph generation with 3 components: 1) Parallel Transformer Encoder, which learns to predict object relationships in a more exclusive manner by grouping relationship labels into groups of similar purpose; 2) Shuffle Transformer, which learns to select the final relationship labels from the category-specific feature generated in the previous step; and 3) Weighted CE loss, used to alleviate the training bias caused by the imbalanced dataset.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑