arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1807.02110cs.CV
收藏

TextTopicNet -通过在语义文本空间上嵌入图像的视觉特征的自监督学习

TextTopicNet - Self-Supervised Learning of Visual Features Through Embedding Images on Semantic Text Spaces

Carnegie Mellon University(卡内基梅隆大学);IIIT Hyderabad(海得拉巴国际信息技术学院);Computer Vision Center, Universitat Autonoma de Barcelona(巴塞罗那自治大学计算机视觉中心);Eurecat, Centre Tecnològic de Catalunya(加泰罗尼亚技术中心 Eurecat)

Yash Patel, Lluis Gomez, Raul Gomez, Marçal Rusiñol, Dimosthenis Karatzas, C. V. Jawahar

摘要:

基于深度学习的方法在计算机视觉中的巨大成功在很大程度上依赖于大规模训练数据集。这些注释丰富的数据集帮助网络学习有区别的视觉特征。收集和注释这些数据集需要大量的人力,并且注释仅限于流行的类集。作为替代方案,通过设计利用免费可用的自我监督的辅助任务来学习视觉特征在计算机视觉社区中变得越来越流行。本文提出了一种利用多模态上下文为计算机视觉算法训练提供自我监督的思想。我们表明,通过训练CNN来预测特定图像更有可能作为插图出现的语义文本上下文,可以有效地学习足够的视觉特征。更具体地说,我们使用流行的文本嵌入技术为深度CNN的训练提供自我监督。我们的实验表明,国家的最先进的性能相比,最近的自我监督或自然监督的方法在图像分类,目标检测,和多模态检索。

英文摘要:

The immense success of deep learning based methods in computer vision heavily relies on large scale training datasets. These richly annotated datasets help the network learn discriminative visual features. Collecting and annotating such datasets requires a tremendous amount of human effort and annotations are limited to popular set of classes. As an alternative, learning visual features by designing auxiliary tasks which make use of freely available self-supervision has become increasingly popular in the computer vision community. In this paper, we put forward an idea to take advantage of multi-modal context to provide self-supervision for the training of computer vision algorithms. We show that adequate visual features can be learned efficiently by training a CNN to predict the semantic textual context in which a particular image is more probable to appear as an illustration. More specifically we use popular text embedding techniques to provide the self-supervision for the training of deep CNN. Our experiments demonstrate state-of-the-art performance in image classification, object detection, and multi-modal retrieval compared to recent self-supervised or naturally-supervised approaches.

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。

↑