2018-05-16

Summary:

Video Caption

[1]《OpenNMT: Open-Source Toolkit for Neural Machine Translation》2017arxiv

Abstract：

arXiv：https://arxiv.org/pdf/1701.02810.pdf

github：https://github.com/zxy14120448/video-caption.pytorch

    https://github.com/zxy14120448/video-caption-openNMT.pytorch

注：

Image Search

《AMC: Attention guided Multi-modal Correlation Learning for Image Search》2017CVPR

关键字：Ranking loss

Abstract:

arxiv: https://arxiv.org/abs/1704.00763

github: https://github.com/kanchen-usc/amc_att

《Class-Weighted Convolutional Features for Visual Instance Search》2017arxiv

关键字：CAM(class activation map),feature aggregation

Abstract:

arxiv: https://arxiv.org/pdf/1707.02581.pdf

github: https://github.com/zxy14120448/retrieval-2017-cam

数据增广的论文

《Greedy AutoAugment》

贪婪的自动数据增广，GAutoAug可提高20.87%的性能（准确率）

链接：https://arxiv.org/pdf/1908.00704.pdf