Large-scale Video Object Segmentation

Workshop in conjunction with ECCV 2024

September 30th, PM, 2024

MiCo Milano, Italy

Introduction

The 6th LSVOS challenge will be held in conjunction with ECCV 2024 in MiCo Milano. In this edition of the workshop and challenge, we replace the classic YouTube-VOS benchmark with MOSE and LVOS to study the VOS under more challenging complex environments. MOSE focuses on complex scenes, including the disappearance-reappearance of objects, inconspicuous small objects, heavy occlusions, crowded environments, etc. LVOS focuses on long-term videos, with complex object motion and long-term reappearance. Besides, we also replace the origin YouTube-RVOS benchmark with MeViS. MeViS focuses on referring the target object in a video through its motion descriptions instead of static attributes, which breaks the basic design principles behind existing RVOS methods and boosts the rethinking of motion modeling. In addition, we will hold a series of talks by the leading experts in video understating.

Dates

Event Date
Challenge release Jul 01, 2024
Validation server online Jul 05, 2024
Test server online Aug 01, 2024
Submission deadline Aug 10, 2024
Notification Aug 15, 2024

Speakers

Nikhila Ravi

Meta AI

Kristen Grauman

University of Texas at Austin

Hengshuang Zhao

The University of Hong Kong

Zongxin Yang

Harvard University

Tracks & Submission

The 6th LSVOS challenge includes two tracks. In this year, we replace the classic YouTube-VOS benchmark with MOSE and LVOS to study the VOS under more challenging complex environments. Besides, we also replace the origin YouTube-RVOS benchmark with MeViS.

Below are the links and task descriptions for the two tracks:

Track1: Video Object Segmentation (VOS)

The video object segmentation task aims to segmenting a particular object instance throughout the entire video sequence given only the object mask of the first frame.

Track 2: Referring Video Object Segmentation (RVOS)

Referring video object segmentation aims to segment an object in video with language expressions.


The final technique report for the two tracks are available now! Please refer to
this link.

Leadboard

Track 1: Video Object Segmentation (VOS)

Rank Team Name Team Members Affiliation J & F | J | F Tech Report
1st PCL VisionLab Deshui Miao1,2, Yameng Gu1, Xin Li2,
Zhenyu He1,2, Yaowei Wang2,
Ming-Hsuan Yang3
1 Harbin Institute of Technology, Shenzhen
2 Peng Cheng Laboratory
3 University of California at Merced
80.90 | 76.16 | 85.63 PDF
2nd yuanjie Jinming Chai, Qin Ma,
Junpei Zhang, Licheng Jiao,
Fang Liu
Intelligent Perception and Image
Understanding Lab, Xidian University
80.84 | 76.42 | 85.26 PDF
3rd Xy-unu Xinyu Liu, Jing Zhang, Kexin Zhang,
Xu Liu, LingLing Li
Intelligent Perception and Image
Understanding Lab, Xidian University
79.52 | 75.16 | 83.88 PDF
4th Sch89.89 Cannot be reached. Cannot be reached. 76.35 | 71.94 | 80.76
4th MVP-TIME Feiyu Pan, Hao Fang,
Runmin Cong, Wei Zhang,
Xiankai Lu
Shandong University 75.79 | 71.25 | 80.33 PDF

Track 2: Referring Video Object Segmentation (RVOS)

Rank Team Name Team Members Affiliation J & F | J | F Tech Report
1st MVP-TIME Hao Fang, Feiyu Pan, Xiankai Lu,
Wei Zhang, Runmin Cong
Shandong University 62.57 | 58.98 | 66.15 PDF
2nd TXT Tuyen Tran Applied Artificial Intelligence Institute, Deakin
University, Australia
60.40 | 57.02 | 63.78 PDF
3rd CASIA_IVA Bin Cao1,2,3, Yisi Zhang4, Hanyi Wang2,
Xingjian He1, Jing Liu1,2
1 Institute of Automation, Chinese Academy
of Sciences,
2 School of Artificial Intelligence, University of
Chinese Academy of Sciences,
3 Beijing Academy of Artificial Intelligence
4 University of Science and Technology Beijing
60.36 | 56.88 | 63.85 PDF

Schedule

Time (GMT+2) Programme (Room: Suite 6)
14:00 - 14:10
Opening Remarks
14:10 - 14:40
Keynote Speaker
Keynote Speaking
Jerome Bell
Nikhila Ravi

Meta AI

14:40 - 14:45
VOS Track Introduction
14:45 - 15:00
Challenge Winners
VOS Winning Teams Talk
PCL VisionLab
15:00 - 15:30
Keynote Speaker
Keynote Speaking
Jerome Bell
Kristen Grauman

University of Texas at Austin

15:30 - 16:00
Coffee Break
16:00 - 16:05
RVOS Track Introduction
16:05 - 16:20
Challenge Winners
RVOS Winning Teams Talk
MVP-TIME
16:20 - 16:50
Keynote Speaker
Keynote Speaking
Jerome Bell
Hengshuang Zhao

The University of Hong Kong

16:50 - 17:20
Keynote Speaker
Keynote Speaking
Jerome Bell
Zongxin Yang

Harvard University

17:20 - 17:30
Award Ceremony and Closing Remarks

Organizers

Lingyi Hong

Fudan University

Henghui Ding

Fudan University

Chang Liu

Nanyang Technological University

Ning Xu

Apple Inc.

Linjie Yang

ByteDance Inc.

Yuchen Fan

Meta Reality Labs

Contact

Feel free to contact us:
henghui.ding@gmail.com
honglyhly@gmail.com