论文阅读笔记（六十六）【ICCV2019】：Adversarial Representation Learning for Text-to-Image Matching
Methodology 作者提出TIMAM (Text-Image Modality Adversarial Matching)方法，比较简洁明了，具体包含三个部分： （1）特征提取器： 文 ...
原文：#ICCV2019论文阅读#Fully_convolutional_geometric_features
一 知识背景 D scan amp cloud points 点云 patch based features，fully convolutional network, deep metric learning, sparse tensors,sparse convolutions, hard negetive mining, contrastive loss, triplet loss, batc ...
2020-01-08 23:01 0 876 推荐指数：
相关推荐
论文阅读笔记五十七:FCOS: Fully Convolutional One-Stage Object Detection(CVPR2019)
论文原址：https://arxiv.org/abs/1904.01355 github: tinyurl.com/FCOSv1 摘要 本文提出了一个基于全卷积的单阶段检测网络，类似于语义分割，针对每个像素进行预测。RetinaNet,SSD,YOLOv3，Faster ...
细粒度语义分割：ICCV2019论文解析
细粒度语义分割：ICCV2019论文解析 Fine-Grained Segmentation Networks: Self-Supervised Segmentation for Improved Long-Term Visual Localization 论文链接 ...
无监督域对抗算法：ICCV2019论文解析
无监督域对抗算法：ICCV2019论文解析 Drop to Adapt: Learning Discriminative Features for Unsupervised Domain Adaptation 论文链接： http ...
噪声标签的负训练：ICCV2019论文解析
噪声标签的负训练：ICCV2019论文解析 NLNL: Negative Learning for Noisy Labels 论文链接： http://openaccess.thecvf.com/content_ICCV_2019/papers ...
人脸真伪验证与识别：ICCV2019论文解析
人脸真伪验证与识别：ICCV2019论文解析 Face Forensics++: Learning to Detect Manipulated Facial Images 论文链接： http://openaccess.thecvf.com/content_ICCV_2019 ...
结构感知图像修复：ICCV2019论文解析
结构感知图像修复：ICCV2019论文解析 StructureFlow: Image Inpainting via Structure-aware Appearance Flow 论文链接： http://openaccess.thecvf.com ...
点云配准的端到端深度神经网络：ICCV2019论文解读
点云配准的端到端深度神经网络：ICCV2019论文解读 DeepVCP: An End-to-End Deep Neural Network for Point Cloud Registration 论文链接： http://openaccess.thecvf.com ...