ECCV 2018 Workshop and Challenge:
Person In Context(PIC)

The Chanllenge Website(www.picdataset.com/challenge/index/ ) is online now!

INTRODUCTION

The 1st Person in Context (PIC) Workshop and Challenge will be held in conjunction with ECCV on the afternoon of September 9th 2018, in Munich Germany. Recent years witness the great progress of the perception task such as image classification, object detection and pixel-wise semantic/instance segmentation. It is the right time to go one step further to infer the relations between the objects. Increasingly more efforts are devoted to relation prediction, such as the Visual Genome and Google Open Image challenge. There are mainly two differences between existing relation prediction works and PIC challenge. Firstly, instead of inferring all the relations between any two objects, PIC focuses on estimating human-centric relations, including human-object relation and human-human relations. Each relation is represented by a triple in the form of <Subject, Relation, Object>, such as <Human A, hold, Bottle A> and <Human A, hug, Human B>. In other words, under the definition of PIC, ‘Subject’ in the triple should be human. Secondly, PIC targets at the relation segmentation. More precisely, traditional relation prediction only estimates the bounding box of ‘Subject’ and ‘Object’ while PIC needs estimate their masks (shapes).

Topics of interest

Multi-person parsing and pose estimation
2D/3D human pose estimations from the single RGB/Depth images or videos
Pedestrian detection in the wild scenarios
Human action recognition and trajectory recognition/prediction
Human re-identification in crowd videos and cross-view cameras
Relationship prediction
Image caption, dense caption
Scene parsing, instance segmentation, semantic segmentation
Novel datasets for performance evaluation and/or empirical analyses of existing methods
Advanced applications of human understanding, including autonomous cars, event recognition and prediction, robotic manipulation, indoor navigation, image/video retrieval and virtual reality.

CHALLENGE

The human-centric dense relation segmentation Challenge

The images are collected from real-world scenarios, with humans appearing with challenging poses and views, heavily occlusions, various appearances and low-resolutions. We provide TBA images with TBA kinds of labels and TBA kinds of relations. For each image, we have TBA instances and TBA relationships on average. In sum, we label TBA instances and TBA relations. We mainly define 2 kinds of relations, including position relation and action relations. Several example images of the dataset are shown in the following.

Awards

1st place

NVIDIA Titan V * 2

2nd place

NVIDIA Titan Xp * 2

3rd place

NVIDIA 1080Ti * 2

SUBMISSION

Paper Submission:: 8 pages (excluding references); LaTex template; LaTex template(camera ready); Submission site: https://cmt3.research.microsoft.com/PICWC2018/
Challenge Submission:: Track: The human-centric relation segmentation Challenge

Important date

Dataset Pre-Release：	June 1st
Challenge Opens:	July 1st
Challenge Deadline:	August 1st
Workshop paper submission deadline:	August ~~4th~~ 12th
Review Due:	August ~~11th~~ 14th
Notification to authors:	August ~~13th~~ 15th
Camera ready deadline:	August ~~15th~~ 17th
Workshop date:	September 9th

WORKSHOP SCHEDULE

Time	Schedule
13:30-13:40	Opening remarks, The Person In Context(PIC) challenge introduction and results
13:40-13:50	Oral talk1: Winner of PIC challenge
13:50-14:00	Oral talk2: Runner-up/third place of PIC challenge
14:00-14:30	Invited talk 1: Bernt Schiele, Max-Planck-Institut für Informatik
14:30-15:00	Invited talk 2: Ming-Hsuan Yang, Professor, University of California at Merced
15:00-15:30	Invited talk 3: Alan Yuille, Professor, Johns Hopkins University
15:30-16:00	Invited talk 4: Jia Deng, Assistant Professor, Princeton University
16:00-16:30	Invited talk 5: Wenjun Zeng, Microsoft Research Asia,
16:30-17:00	Invited talk 6: Tao Mei, Technical Vice President of JD.com
17:00-17:10	Awards & Future Plans