上海科技大学信息科学与技术研讨会（SSIST day1) 笔记

一、Deep representations，Adversial learning and Domain Adaptation for Some Computer Vision Problems

Presenters:Rama Chellappa University of Maryland
1.Deep representations:
Learning Deep Representations of Fine-Grained Visual Descriptions(CVPR2016)
本文提出了一个模型，可以用来做什么呢？就是给你一句话，搜出满足这句话的图像。可参考： https://blog.csdn.net/cv_family_z/article/details/52061403?locationNum=10
2.Adversial learning:
主要介绍了UMD face analytics system的训练、网络结构和结果，具体可参考：https://www.umd.edu/
原文：Deep intermodal video analytics
链接：https://www.iarpa.gov/index.php/research-programs/diva
3.Domain Adaptation：
可参考：https://blog.csdn.net/matafela/article/details/77827217
听报告的时候拍了几张照片，手抖，勉强看一下还是可以的

二、Visual Domain Adaptation

Presenters:Dong Xu The University of Sydney
1.Domain Adversarial Neural Network(DANN):
原文：Domain-Adversarial Training of Neural Networks（JMLR 2016, ICML 2015）
链接：https://arxiv.org/abs/1505.07818
2.Collaborative and Adversarial Network(CAN)，Incremental CAN(iCAN):
原文：Collaborative and Adversarial Network for Unsupervised Domain Adaptation（CVPR 2018）

三、3D Visual Reconstruction of Non-Rigid Dynamic Scenes

Presenters: Hongdong Li Australian National University
1.Superpixel Soup Algorithm:Dense Complex Dynamic Scene Reconstruction from a monocular camera
原文：Monocular Dense 3D Reconstruction of a Complex Dynamic Scene from Two Perspective Frames（ICCV 2017）
链接：https://arxiv.org/abs/1708.04398
2. SFRM: Structure from Recurrent Motion for 3D human pose recovery
原文：Structure from Recurrent Motion: From Rigidity to Recurrency（CVPR 2018）
链接：https://arxiv.org/abs/1804.06510

四、Bridging Visual-Semantic Gap via Deep Multimodal Learning

Presenters: Liang Wang Chinese Academy of Sciences

五、Deep Learning in Medical Image Synthesis,Quantification, and Prediction

Presenters: Dinggang Shen United Imaging Intelligence

六、Recent Development in Image Generation and Processing

Presenters: Jiaya Jia Tencent
1.Super-resolution
原文：Detail-revealing Deep Video Super-resolution（ICCV 2017)
链接：https://arxiv.org/abs/1704.02738
2. Deblurring
Scale-recurrent Network for Deep Image Deblurring(CVPR 2018)
3.Image Generation
Semi-parametric Image Synthesis(CVPR 2018)