CVPR 2021 best paper candidates 목록

저번 글에서 이번 CVPR 베스트 페이퍼를 살펴봤는데,

CVPR 2021 Best paper는 아니지만 후보에 올랐던 논문들을 한번씩 보려고 한다.

 

283	Privacy-Preserving Image Features via Adversarial Affine Subspace Embeddings	Mihai Dusmanu (ETH Zurich); Johannes L Schönberger (Microsoft); Sudipta Sinha (Microsoft); Marc Pollefeys (ETH Zurich / Microsoft)
415	Learning Calibrated Medical Image Segmentation via Multi-Rater Agreement Modeling	Wei Ji (University of Alberta); Shuang Yu (Tencent); Junde Wu (Harbin Institute of Technology); Kai Ma (Tencent); Cheng Bian (Tencent); Qi Bi (University of Amsterdam);
456	Diffusion Probabilistic Models for 3D Point Cloud Generation	Shitong Luo (Peking University); Wei Hu (Peking University)
566	Task Programming: Learning Data Efficient Behavior Representations	Jennifer J. Sun (Caltech); Ann Kennedy (Northwestern University); Eric Zhan (Caltech); David J. Anderson (Caltech); Yisong Yue (Caltech); Pietro Perona (California Institute of Technology)
902	PoseAug: A Differentiable Pose Augmentation Framework for 3D Human Pose Estimation	Kehong Gong (National University of Singapore); Jianfeng Zhang (NUS); Jiashi Feng (NUS)
1058	SCANimate: Weakly Supervised Learning of Skinned Clothed Avatar Networks	Shunsuke Saito (Facebook); Jinlong Yang (Max Planck Institute for Intelligent Systems); Qianli Ma (Max Planck Institute for Intelligent Systems); Michael J. Black (Max Planck Institute for Intelligent Systems)
1078	On Self-Contact and Human Pose	Lea Müller (Max Planck Institute for Intelligent Systems); Ahmed A A Osman (Max Planck Institute for Intelligent Systems); Siyu Tang (ETH Zurich); Chun-Hao Paul Huang (Max Planck Institute for Intelligent Systems); Michael J. Black (Max Planck Institute for Intelligent Systems)
1269	Binary TTC: A Temporal Geofence for Autonomous Navigation	Abhishek Badki (University of California, Santa Barbara); Orazio Gallo (NVIDIA Research); Jan Kautz (NVIDIA); Pradeep Sen (UC Santa Barbara)
1300	Rethinking and Improving the Robustness of Image Style Transfer	Pei Wang (UC San Diego); Yijun Li (Adobe Research); Nuno Vasconcelos (UC San Diego)
1704	Audio-Visual Instance Discrimination with Cross-Modal Agreement	Pedro Morgado (University of California, San Diego); Nuno Vasconcelos (UCSD, USA); Ishan Misra (Facebook AI Research)
1824	Point2Skeleton: Learning Skeletal Representations from Point Clouds	Cheng Lin (The University of Hong Kong); Changjian Li (University College London); Yuan Liu (The University of Hong Kong); Nenglun Chen (The University of Hong Kong); Yi King Choi (The University of Hong Kong); Wenping Wang (The University of Hong Kong)
1929	Human POSEitioning System (HPS): 3D Human Pose Estimation and Self-Localization in Large Scenes From Body-Mounted Sensors	Vladimir Guzov (Max Planck Institute for Informatics); Aymen Mir (Max Planck Institute of Informatics); Torsten Sattler (Czech Technical University in Prague); Gerard Pons-Moll (MPII, Germany)
2551	Where and What? Examining Interpretable Disentangled Representations	Xinqi Zhu (University of Sydney); Chang Xu (University of Sydney); Dacheng Tao (The University of Sydney)
	https://arxiv.org/pdf/2104.05622.pdf
3225	Learning To Recover 3D Scene Shape From a Single Image	Wei Yin (University of Adelaide); Jianming Zhang (Adobe Research); Oliver Wang (Adobe Systems Inc); Simon Niklaus (Adobe Research); Long T Mai (Adobe Research); Simon Chen (Adobe Research); Chunhua Shen (University of Adelaide)
3367	GIRAFFE: Representing Scenes As Compositional Generative Neural Feature Fields	Michael Niemeyer (Max Planck Institute for Intelligent Systems, Tübingen and University of Tübingen); Andreas Geiger (MPI-IS and University of Tuebingen)
		이게 베스트 페이퍼. 저번에 리뷰 했으니 스킵
3386	Polygonal Building Extraction by Frame Field Learning	Nicolas Girard (Inria Sophia-Antipolis); Dmitriy Smirnov (MIT); Justin M Solomon (MIT); Yuliya Tarabalka (Inria Sophia-Antipolis)
	https://arxiv.org/pdf/2004.14875.pdf	segmentation결과를 실제 Downstream task들에 사용할때를 고려해서 building들을 뽑는 것 같다
3433	NeuralRecon: Real-Time Coherent 3D Reconstruction From Monocular Video	Jiaming Sun (SenseTime); Yiming Xie (SenseTime); Linghao Chen (Zhejiang University); Xiaowei Zhou (Zhejiang University); Hujun Bao (Zhejiang University)
3592	CoCosNet v2: Full-Resolution Correspondence Learning for Image Translation	Xingran Zhou (Zhejiang University); Bo Zhang (Microsoft Research Asia); Ting Zhang (MSRA); Pan Zhang (USTC); Jianmin Bao (Microsoft Research Asia); Dong Chen (Microsoft Research Asia); Zhongfei Zhang (Binghamton University); Fang Wen (Microsoft Research Asia)
	https://openaccess.thecvf.com/content/CVPR2021/papers/Zhou_CoCosNet_v2_Full-Resolution_Correspondence_Learning_for_Image_Translation_CVPR_2021_paper.pdf	image to image translation을 잘 하는 논문인 것 같다. patch match기법을 가지고 점점 맞춘다는데...
4263	Less Is More: ClipBERT for Video-and-Language Learning via Sparse Sampling	Jie Lei (UNC Chapel Hill); Linjie Li (Microsoft); Luowei Zhou (Microsoft); Zhe Gan (Microsoft); Tamara Berg (UNC Chapel Hill, USA); Mohit Bansal (University of North Carolina at Chapel Hill); Jingjing Liu (Microsoft)
	https://arxiv.org/pdf/2102.06183.pdf	VQA처럼 비전이랑 NLP 같이해야하는 쪽에서 많이 언급되었던 논문으로 아는데, NLP 관심 없어서 스킵...
4286	Neural Body: Implicit Neural Representations With Structured Latent Codes for Novel View Synthesis of Dynamic Humans	Sida Peng (Zhejiang University); Yuanqing Zhang (Zhejiang University); Yinghao Xu (Chinese University of Hong Kong); Qianqian Wang (Cornell); Qing Shuai (Zhejiang University); Hujun Bao (Zhejiang University); Xiaowei Zhou (Zhejiang University)
	https://arxiv.org/pdf/2012.15838.pdf	일단 view 합성하는 분야. ill-posed problem을 해결하기 위해서, 비디오 프레임들이 모두 공유하는 'structured' latent code만든다고 한다. Neural body라는걸 제안하는데, 이게 사람 몸에 대한 neural representation을 구조적으로 배운다. 대략 살펴보니까 color나 density를 모델이 있는 것 같은데...
4418	Exploring Simple Siamese Representation Learning	Xinlei Chen (FAIR); Kaiming He (Facebook AI Research)
		이건 다음번에 팀원분이 발표해주신다해서 스킵
4551	Guided Interactive Video Object Segmentation Using Reliability-Based Attention Maps	Yuk Heo (Korea University); Yeong Jun Koh (Chungnam National University); Chang-Su Kim (Korea university)
4877	GeoSim: Realistic Video Simulation via Geometry-Aware Composition for Self-Driving	Yun Chen (Uber ATG); Frieda Rong (Uber ATG); Shivam Duggal (Delhi Technological University); Shenlong Wang (Uber ATG, University of Toronto); Xinchen Yan (Uber ATG); Sivabalan Manivasagam (University of Toronto); Shangjie Xue (MIT); Ersin Yumer (Uber ATG); Raquel Urtasun (Uber ATG)
4945	Neural Lumigraph Rendering	Petr Kellnhofer (Stanford University); Lars C Jebe (Raxium); Andrew Jones (Raxium); Ryan Spicer (Raxium); Kari Pulli (University of Oulu); Gordon Wetzstein (Stanford University)
5291	Event-Based Synthetic Aperture Imaging With a Hybrid Network	Xiang Zhang (Wuhan University); Wei Liao (WuHan University); Lei Yu (Wuhan University); Wen Yang (Wuhan University); Gui-Song Xia (Wuhan University)
5562	Energy-Based Learning for Scene Graph Generation	Mohammed Suhail (University of British Columbia); Abhay Mittal (Amazon); Behjat Siddiquie (Amazon); Christopher Broaddus (Amazon); Jayan Eledath (Amazon); gerard medioni (USC); Leonid Sigal (University of British Columbia)
6333	Learning High Fidelity Depths of Dressed Humans by Watching Social Media Dance Videos	Yasamin Jafarian (University of Minnesota); Hyun Soo Park (The University of Minnesota)
7124	MP3: A Unified Model To Map, Perceive, Predict and Plan	Sergio Casas (Uber ATG / University of Toronto); Abbas Sadat (Uber ATG); Raquel Urtasun (Uber ATG)
8458	NeX: Real-Time View Synthesis With Neural Basis Expansion	Suttisak Wizadwongsa (Vidyasirimedhi Institute of Science and Technology); Pakkapon Phongthawee (Vidyasirimedhi Institute of Science and Technology); Jiraphon Yenphraphai (Vidyasirimedhi Institute of Science and Technology); Supasorn Suwajanakorn (Vidyasirimedhi Institute of Science and Technology)
8866	NewtonianVAE: Proportional Control and Goal Identification From Pixels via Physical Latent Spaces	Miguel Jaques (University of Edinburgh); Michael Burke (Monash University); Timothy Hospedales (Edinburgh University)
	https://openaccess.thecvf.com/content/CVPR2021/papers/Jaques_NewtonianVAE_Proportional_Control_and_Goal_Identification_From_Pixels_via_Physical_CVPR_2021_paper.pdf	뭔가 control가능한 latent space를 학습한다는 것 같은데 글이 안읽힌다 ㄱ-
10237	Fast End-to-End Learning on Protein Surfaces	Freyr Sverrisson (EPFL); Jean Feydy (Imperial College London); Bruno Correia (EPFL); Michael Bronstein (Imperial College London / Twitter)
10509	Real-Time High-Resolution Background Matting	Shanchuan Lin (University of Washington); Andrey Ryabtsev (University of Washington); Soumyadip Sengupta (University of Washington); Brian Curless (University of Washington); Steve Seitz (University of Washington); Ira Kemelmacher-Shlizerman (University of Washington)