ICCV 2019 行为识别/视频理解论文汇总
加入极市专业CV交流群,与6000+来自腾讯,华为,百度,北大,清华,中科院等名企名校视觉开发者互动交流!更有机会与李开复老师等大牛群内互动!
同时提供每月大咖直播分享、真实项目需求对接、干货资讯汇总,行业技术交流。关注 极市平台 公众号 ,回复 加群,立刻申请入群~
作者:木石
https://zhuanlan.zhihu.com/p/77742840
来源:知乎,已获作者授权转载,禁止二次转载。
前两天稍微整理了下 ICCV 2019 已经放出来的视频理解/行为识别相关论文,这里分享下。
Graph Convolutional Networks for Temporal Action Localization
作者:Chuang Gan 等
Action recognition with spatial-temporal discriminative filter banks
作者:Yuanjun Xiong 等
AssembleNet: Searching for Multi-Stream Neural Connectivity in Video Architectures
作者:Google Brain
neural architecture search for video understanding——大力出奇迹
DynamoNet: Dynamic Action and Motion Network
作者:Ali Diba Luc Van Gool
Reasoning About Human-Object Interactions Through Dual Attention Networks
作者:Bolei Zhou
Learning Temporal Action Proposals with Fewer Labels
作者:Stanford Feifei组 Juan Carlos Niebles
EPIC-Fusion: Audio-Visual Temporal Binding for Egocentric Action Recognition
作者:Dima Damen 等
SlowFast Networks for Video Recognition
(文章链接:https://arxiv.org/abs/1812.03982)
kaiming 大神 from FAIR
Video Classification with Channel-Separated Convolutional Networks
(文章链接:https://arxiv.org/abs/1904.02811)
Du Tran 大神 from FAIR
SCSampler: Sampling Salient Clips from Video for Efficient Action Recognition. oral
(文章链接:https://arxiv.org/abs/1904.04289)
Du Tran 大神 from FAIR
DistInit: Learning Video Representations without a Single Labeled Video.
(文章链接:https://arxiv.org/abs/1901.09244)
Du Tran 大神 from FAIR
很简单的思路
TSM: Temporal Shift Module for Efficient Video Understanding
作者:Ji Lin, Chuang Gan, Song Han
论文链接:https://arxiv.org/abs/1811.08383
Github链接:https://github.com/mit-han-lab/temporal-shift-module
emmm感觉吧,就像是搞了个带Mask的固定卷积核?
BMN: Boundary-Matching Network for Temporal Action Proposal Generation
(文章链接:https://arxiv.org/abs/1907.09702)
来自作者大大解读:林天威:[ICCV 2019][时序动作提名] 边界匹配网络详解
(原文链接:https://zhuanlan.zhihu.com/p/75444151)
Weakly Supervised Energy-Based Learning for Action Segmentation.oral
文章链接:https://github.com/JunLi-Galios/CDFL
Pose-aware Dynamic Attention for Human Object Interaction Detection
文章链接:https://github.com/bobwan1995/PMFNet
What Would You Expect? Anticipating Egocentric Actions With Rolling-Unrolling LSTMs and Modality Attention
项目链接:https://iplab.dmi.unict.it/rulstm/
论文链接:https://arxiv.org/pdf/1905.09035.pdf
GitHub:https://github.com/fpv-iplab/rulstm
Fine-Grained Action Retrieval Through Multiple Parts-of-Speech Embeddings
论文链接:https://arxiv.org/abs/1908.03477
项目链接:https://mwray.github.io/FGAR/
HowTo100M: Learning a Text-Video Embedding by Watching Hundred Million Narrated Video Clips
作者:Antoine Miech, Dimitri Zhukov, Jean-Baptiste Alayrac, Makarand Tapaswi, Ivan Laptev, Josef Sivic
论文链接:https://arxiv.org/abs/1906.03327
项目链接:https://github.com/antoine77340/howto100m
code(链接:https://github.com/antoine77340/howto100m)
Temporal Attentive Alignment for Large-Scale Video Domain Adaptation
作者:Min-Hung Chen, Zsolt Kira, Ghassan AlRegib, Jaekwon Woo, Ruxin Chen, Jian Zheng
论文链接:https://arxiv.org/abs/1907.12743
Github链接:https://github.com/cmhungsteve/TA3N
STM- SpatioTemporal and Motion Encoding for Action Recognition
from ZJU && SenseTime Group Limited
论文链接:https://arxiv.org/abs/1908.02486
PS. 附上我们的ICCV2019GitHub开源项目:
https://github.com/extreme-assistant/iccv2019
-End-
*延伸阅读
添加极市小助手微信(ID : cv-mart),备注:研究方向-姓名-学校/公司-城市(如:目标检测-小极-北大-深圳),即可申请加入目标检测、目标跟踪、人脸、工业检测、医学影像、三维&SLAM、图像分割等极市技术交流群,更有每月大咖直播分享、真实项目需求对接、干货资讯汇总,行业技术交流,一起来让思想之光照的更远吧~
△长按添加极市小助手
△长按关注极市平台
觉得有用麻烦给个在看啦~