人工智能培训

搜索

人工智能论文:高光谱数据增强(Hyperspectral Data Augmentation)

[复制链接]
usjerseys889 发表于 2019-3-15 12:45:20 | 显示全部楼层 |阅读模式
usjerseys889 2019-3-15 12:45:20 206 0 显示全部楼层
人工智能论文:高光谱数据增强(Hyperspectral Data Augmentation)数据增强是一种流行的技术,有助于提高深度神经网络的泛化能力。它在远程感知场景中发挥着关键作用,其中高质量的地面实况数据的数量有限,并且获取新的例子是昂贵的或不可能的。这是高光谱成像中的常见问题,其中图像数据的手动注释是困难的,昂贵的并且易于人为偏见。在这封信中,我们提出了在深度网络训练之前在推理期间执行的高光谱数据的在线数据增强。这与其他最先进的高光谱增强算法形成对比,后者增加了训练集的大小(和代表性)。此外,我们引入了基于增量的新主成分分析。实验表明,我们的数据增强算法改善了深度网络的泛化,实时工作,并且在线方法可以有效地与离线技术相结合,以提高分类准确性。
Data augmentation is a popular technique which helps improve generalizationcapabilities of deep neural networks.It plays a pivotal role in remote-sensingscenarios in which the amount of high-quality ground truth data is limited, andacquiring new examples is costly or impossible.This is a common problem inhyperspectral imaging, where manual annotation of image data is difficult,expensive, and prone to human bias.In this letter, we propose online dataaugmentation of hyperspectral data which is executed during the inferencerather than before the training of deep networks.This is in contrast to allother state-of-the-art hyperspectral augmentation algorithms which increase thesize (and representativeness) of training sets.Additionally, we introduce anew principal component analysis based augmentation.The experiments revealedthat our data augmentation algorithms improve generalization of deep networks,work in real-time, and the online approach can be effectively combined withoffline techniques to enhance the classification accuracy.人工智能论文:高光谱数据增强(Hyperspectral Data Augmentation) gj20PJyyCy2seHv2.jpg
URL地址:https://arxiv.org/abs/1903.05580     ----pdf下载地址:https://arxiv.org/pdf/1903.05580    ----人工智能论文:高光谱数据增强(Hyperspectral Data Augmentation)
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则 返回列表 发新帖

usjerseys889当前离线
新手上路

查看:206 | 回复:0

快速回复 返回顶部 返回列表