UMD Researchers Present 26 Papers at 2023 Computer Vision and Pattern Recognition Conference

The University of Maryland will present 26 papers at the annual IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2023. Scheduled to take place June 18-22 in Vancouver, Canada, the conference serves as a global platform for leading experts, researchers and industry professionals in the computer vision and pattern recognition community. Participation by UMD computer scientists promises to stimulate meaningful discussions, foster new collaborations and inspire future breakthroughs in the field.

"Having 26 accepted papers from our department for the conference is an exceptional accomplishment that truly reflects the expertise and dedication of our faculty and researchers in Computer Vision," said Department Chair Matthias Zwicker, who holds the Elizabeth Iribe Chair for Innovation and the Phillip H. and Catherine C. Horvitz Professorship. " These accepted papers are a testament to the high-quality research conducted at our institution, and we look forward to sharing our groundbreaking findings with the wider scientific community at the conference." 

CVPR ranks among the leading conferences in the computer vision and machine learning field. Google Scholar metrics place it as one of the top four publications across all categories, surpassed only by Nature, The New England Journal of Medicine, and Science.

The accepted papers are the following:

Learning Expressive Prompting With Residuals for Vision Transformers 

Rajshekhar Das (Carnegie Mellon University) · Yonatan Dukler (AWS AI) · Avinash Ravichandran (Cruise) · Ashwin Swaminathan (UMD)

Progressively Optimized Local Radiance Fields for Robust View Synthesis 

Andréas Meuleman (Korea Advanced Institute of Science and Technology) · Yu-Lun Liu (National Yang Ming Chiao Tung University) ·  Chen Gao (Meta) · Jia-Bin Huang (UMD) · Changil Kim (Facebook) · Min H. Kim (KAIST) · Johannes Kopf (Facebook)

TMO: Textured Mesh Acquisition of Objects With a Mobile Device by Using Differentiable Rendering 

Jaehoon Choi (UMD) · Dongki Jung (Naver Labs) · Taejae Lee (NaverLabs) · Sangwook Kim (Naver Labs) · Youngdong Jung (Naver Labs) · Dinesh Manocha (UMD) · Donghwan Lee (Naver Labs)

FlexNeRF: Photorealistic Free-Viewpoint Rendering of Moving Humans From Sparse Views

Vinoj Jayasundara (UMD) · Amit Agrawal (Amazon) · Nicolas Heron (Amazon) · Abhinav Shrivastava (UMD) · Larry S. Davis (Amazon)

CUDA: Convolution-Based Unlearnable Datasets 

Vinu Sankar Sadasivan (UMD) · Soltanolkotabi (University of Southern California) · Soheil Feizi (UMD)

HaLP: Hallucinating Latent Positives for Skeleton-Based Self-Supervised Learning of Actions 

Anshul Shah (Johns Hopkins University) · Aniket Roy (Johns Hopkins University) · Ketul Shah (Johns Hopkins University) · Shlok Mishra (None) · David Jacobs (UMD) · Anoop Cherian (None) · Rama Chellappa (Johns Hopkins University)

PAniC-3D: Stylized Single-View 3D Reconstruction From Portraits of Anime Characters 

Shuhong Chen (UMD) · Kevin Zhang (UMD) · Yichun Shi (ByteDance) · Heng Wang (Bytedance) · Yiheng Zhu (ByteDance) · Guoxian Song (Bytedance Inc) · Sizhe An (University of Wisconsin-Madison) · Janus Kristjansson (UMD) · Xiao Yang (Bytedance) · Matthias Zwicker (UMD)

Progressive Transformation Learning for Leveraging Virtual Images in Training 

Yi-Ting Shen (UMD ECE) · Hyungtae Lee (DEVCOM Army Research Laboratory) · Heesung Kwon (DEVCOM Army Research Laboratory) · Shuvra S. Bhattacharyya (UMD)

SimpSON: Simplifying Photo Cleanup With Single-Click Distracting Object Segmentation Network 

Chuong Huynh (UMD) · Yuqian Zhou (University of Illinois, Urbana-Champaign) · Zhe Lin (Adobe Research) · Connelly Barnes (Adobe Systems) · Eli Shechtman (Adobe) · Sohrab Amirghodsi (Adobe) · Abhinav Shrivastava (UMD)

DC2: Dual-Camera Defocus Control by Learning To Refocus 

Hadi Alzayer (UMD) · Abdullah Abuolaim (Google) · Leung Chun Chan (University of California, San Diego) · Yang Yang (Google) · Ying Chen Lou (Google) · Jia-Bin Huang (UMD) · Abhishek Kar (Google)

HyperReel: High-Fidelity 6-DoF Video With Ray-Conditioned Sampling 

Benjamin Attal (CMU, Carnegie Mellon University) · Jia-Bin Huang (UMD) · Christian Richardt (Meta Reality Labs) · Michael Zollhöfer (Facebook) · Johannes Kopf (Facebook) · Matthew O’Toole (Carnegie Mellon University) · Changil Kim (Facebook)

Align and Attend: Multimodal Summarization With Dual Contrastive Losses

Bo He (UMD) · Jun Wang (UMD) · Jielin Qiu (Carnegie Mellon University) · Trung Bui (Adobe Research) · Abhinav Shrivastava (UMD) · Zhaowen Wang (Adobe Research)

Towards Scalable Neural Representation for Diverse Videos

Bo He (UMD) · Xitong Yang (Meta) · Hanyu Wang (UMD) · Zuxuan Wu (Fudan University) · Hao Chen (UMD) · Shuaiyi Huang (UMD) · Yixuan Ren (UMD) · Ser-Nam Lim (Meta AI) · Abhinav Shrivastava (UMD)

Shape-Aware Text-Driven Layered Video Editing

Yao-Chih Lee (UMD) · Ji-Ze Genevieve Jang (UMD) · Yi-Ting Chen (None) · Elizabeth Qiu (UMD) · Jia-Bin Huang (UMD)

Consistent View Synthesis With Pose-Guided Diffusion Models

Hung-Yu Tseng (Meta) · Qinbo Li (Facebook) · Changil Kim (Facebook) · Suhib Alsisan (Meta) · Jia-Bin Huang (UMD) · Johannes Kopf (Facebook)

Teaching Matters: Investigating the Role of Supervision in Vision Transformers 

Matthew Walmer (UMD) · Saksham Suri (UMD) · Kamal Gupta (UMD) · Abhinav Shrivastava (UMD)

AdaMAE: Adaptive Masking for Efficient Spatiotemporal Learning With Masked Autoencoders 

Wele Gedara Chaminda Bandara (Johns Hopkins University) · Naman Patel (Zippin) · Ali Gholami (Simon Fraser University) · Mehdi Nikkhah (Zippin) · Motilal Agrawal (UMD) · Vishal M. Patel (Johns Hopkins University)

NIRVANA: Neural Implicit Representations of Videos With Adaptive Networks and Autoregressive Patch-Wise Modeling

Shishira R. Maiya (UMD) · Sharath Girish (UMD) · Max Ehrlich (UMD) · Hanyu Wang (UMD rk) · Kwot Sin Lee (Snap Inc.) · Patrick Poirson (Snap Inc.) · Pengxiang Wu (Snap Inc.) · Chen Wang (Snap Inc.) · Abhinav Shrivastava (UMD)

—by Samuel Malede Zewdu, CS Communications. List of papers and associated names provided by Jia-Bin Huang