Leopold's Blog
  文章分类
数据结构与算法 11 计算机视觉 6 深度学习 1
arxiv 2020 NAS-adversarial attacks arxiv 2020 NAS-adversarial attacks
一篇还挂在arxiv上的很有意思的神经网络架构搜索(NAS)论文链接-自解释用于神经网络架构搜索 这里主要介绍它的思路 首先总结文章的主要思想：通过搜索一个解释网络A“更好的解释目标任务”，解释的准则是通过对抗攻击样本学一个像素位置权重$\
2020-12-24 计算机视觉
NAS Adversarial Attacks
CVPR2020 Geometric Adversarial Attacks and Defenses on 3D Point Clouds CVPR2020 Geometric Adversarial Attacks and Defenses on 3D Point Clouds
CVPR2020 Geometric Adversarial Attacks and Defenses on 3D Point Clouds摘要深度神经网络很容易出现恶意攻击，这些攻击会恶意更改网络的结果。在这项工作中，我们从几何角度探索对
2020-12-14 计算机视觉
对抗攻击 点云数据
HRNet-OCR by Pytorch HRNet-OCR by Pytorch
HRNet-OCR的原代码将各种超参数和训练超参数冗杂在一起,导致不用于其他数据集会很不方便,为了方便调用模型,对模型的超参进行抽离后得到简易版模型实现。 OCR如下图 # ---------------------------------
2020-12-09 计算机视觉
语义分割 SOTA
CVPR2018-Learning to Adapt Structured Output Space for Semantic Segmentation CVPR2018-Learning to Adapt Structured Output Space for Semantic Segmentation
论文链接As the labeling process is tedious and labor intensive, developing algorithms that can adapt source ground truth l
2020-12-04 计算机视觉
语义分割 域自适应
CVPR2020-GraphTER-Unsupervised Learning of Graph Transformation Equivariant Representations via Auto-Encoding Node-wise Transformations CVPR2020-GraphTER-Unsupervised Learning of Graph Transformation Equivariant Representations via Auto-Encoding Node-wise Transformations
Recent advances in Graph Convolutional Neural Networks (GCNNs) have shown their efficiency for nonEuclidean data on gr
2020-12-04 计算机视觉
点云数据 无监督 图卷积
CVPR2020-Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery CVPR2020-Foreground-Aware Relation Network for Geospatial Object Segmentation in High Spatial Resolution Remote Sensing Imagery
These general semantic segmentation methods mainly focus on multi-scale context modeling, ignoring the special issues
2020-12-04 计算机视觉
语义分割 遥感图像