arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1801.09718cs.CV
收藏

VQA中基于对象的推理

Object-based reasoning in VQA

IBM Research, Almaden(IBM研究院阿尔马登分院);University of Chicago(芝加哥大学)

Mikyas T. Desta, Larry Chen, Tomasz Kornuta

摘要:

视觉问答是一个新的问题领域,其中必须处理多模态输入,以解决以自然语言形式给出的任务。由于解决方案本质上需要将视觉和自然语言处理与抽象推理相结合,因此该问题被认为是AI完全的。最近的进展表明,使用从输入中提取的高级抽象事实可能有助于推理。按照这个方向,我们决定开发一种结合最先进的目标检测和推理模块的解决方案。在平衡良好的CLEVR数据集上取得的结果证实了承诺,并显示出复杂的“计数”任务的准确性有了显著的提高。

英文摘要:

Visual Question Answering (VQA) is a novel problem domain where multi-modal inputs must be processed in order to solve the task given in the form of a natural language. As the solutions inherently require to combine visual and natural language processing with abstract reasoning, the problem is considered as AI-complete. Recent advances indicate that using high-level, abstract facts extracted from the inputs might facilitate reasoning. Following that direction we decided to develop a solution combining state-of-the-art object detection and reasoning modules. The results, achieved on the well-balanced CLEVR dataset, confirm the promises and show significant, few percent improvements of accuracy on the complex "counting" task.

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。

↑