Munich, Germany, 8th September
2018, in conjunction with
ECCV'18
About
In contrast to humans that are
very good in anticipating
the behavior of other objects,
animals, or humans, developing
methods that anticipate
human behavior from video
or other sensor data is
very challenging and has
just recently received
an increase of interest.
In the past, the features
for analyzing, in particular,
visual data like images
or videos were too weak
such that approaches that
predict the future were
unlikely to succeed.
This burden has been overcome
due to recent progress
in this field. The anticipation
of human behavior, however,
is not well defined in
the literature and varies
depending on the task in
terms of granularity and
time horizon. In the context
of driver assistance systems,
the prediction of the trajectory
of a pedestrian needs to
be within centimeter accuracy
but only for a very short
time horizon of one second.
For tracking applications
or motion planning, the
potential destination of
a human and trajectories
of several seconds or minutes
to reach the destination
need to be predicted. In
order to prioritize several
tasks for a service robot
during a day, only the
rough time and location
of an activity is needed.
For instance, when the
robot anticipates that
the owner wants to cook
in one hour, the robot
will be in the kitchen
at the right time.
The purpose of this workshop is
to discuss recent approaches
that anticipate human behavior
from video or other sensor
data, to bring together
researchers from multiple
fields and perspectives,
and to discuss major research
problems and opportunities
and how we should coordinate
efforts to advance the
field.
The topics of interest
for this workshop include,
but are not limited to:
Early activity recognition
Anticipation of trajectories
Anticipation of human
poses
Anticipation of activities
or events
Anticipation of group
behavior
Predicting frames, features
or semantic in videos
or other sensor data
Predicting use of objects
or affordances
Datasets, evaluation,
and benchmarking
Applications including
but not limited to robotics,
autonomous systems, virtual/augmented
reality
Prospective authors will be invited
to submit a regular paper
of previously unpublished
work (ECCV workshop format)
or an extended abstract
of a published or ongoing
work via the workshop webpage.
All the submissions will
be reviewed by the workshop’s
international program committee.
Accepted regular papers
will be presented during
the oral or poster sessions
and included in the ECCV
workshop proceedings. Accepted
extended abstracts will
be presented at the poster
session.
We provide a limited number
of travel grants for early
career researchers, which
present their work at the
workshop.
The workshop will be located in room N1080ZG at TU München,
Arcisstraße 21 The posters will be located in room 2607 and 3607 (upstairs gallery of 2607).
Prospective authors are invited
to submit a 14-page paper
(ECCV format) via the workshop
submission web-page. The
papers will be reviewed
by an international Program
Committee. Reviews will
be double-blind according
to ECCV guidelines. The
papers are expected to present
novel work.
Authors should take into account
the following: All papers
must be written in English
and submitted in PDF format.
The maximum paper length
is 14 pages, excluding references.
The workshop paper submission
guidelines are the same
as the Main Conference papers.
Submissions will be rejected
without review if they:
contain more than 14 pages
(excluding references);
violate the double-blind
policy; violate the dual-submission
policy. Authors will have
the opportunity to submit
supplementary material.
Accepted papers will be published
by Springer together with
all other ECCV workshop
papers as post-proceedings.
Regular workshop publications
are intended to be treated
just like conference publications,
in that at least one of
the authors is required
to register for the conference
(and hence pay the registration
fee).
Extended Abstracts
Prospective authors are invited
to submit a 2-page extended
abstract (ECCV format) via
the workshop submission
web-page. The abstracts
will be reviewed by an international
Program Committee. Reviews
will be single-blind according
to ECCV guidelines. The
extended abstracts might
summarize already published
work, work under review,
or work in progress.
Authors should take into account
the following: All papers
must be written in English
and submitted in PDF format.
The maximum paper length
is 2 pages, excluding references.
In contrast to regular papers,
the authors names should
be added to the extended
abstract.
Accepted extended abstract will
not be published and dot
not count as a publication.
Instead, the authors will
present their work either
as poster or oral presentation
at the workshop. Registration
for the workshop will be
required.
If you wish to be considered for
a travel grant, please
mark 'yes' to the question
'Are you applying for a
travel grant?' when submitting
your paper or extended
abstract on CMT.
We provide a limited number of
travel grants for early
career researchers, which
present their work at the
workshop. The travel grants
will cover parts of the
travel cost (conference
fee, accommodation, transport).
The amount (about 400 EUR)
will be determined. The
selection will be based
on the quality of the submission
and the current academic
career stage. We aim to
support in particular female
researchers.
If you wish to be considered for
a travel grant, please mark
'yes' to the question 'Are
you applying for a travel
grant?' when submitting
your paper or extended abstract
on CMT. The grant will be
awarded to the presenter
who is required to be co-author
of the submission but not
necessary first author.
Eligible are students at
Master or PhD level and
PostDocs at universities
and academic institutions.
It is allowed to apply for
other travel grants as well.
If you have been awarded a travel
grant, you will get detailed
instructions per email.
Only actual expenses will
be reimbursed up to the
amount of the grant. For
this, you have to send us
the original receipts after
the workshop. If you receive
a travel grant from another
source, you can use both
as long as the travel grants
reimburse different expenses.
For instance, you can use
one grant for the conference
fee and the second for the
accommodation.
Speakers
Prof. Michael S. Ryoo
is an Assistant Professor in the Department of Computer
Science at Indiana University Bloomington. His research interest is within
the areas of Computer Vision and Robotics, with a particular emphasis on
human activity recognition/learning, first-person vision, and robot action
learning. Before joining IU, Dr. Ryoo was a staff researcher at the
Robotics Section of the NASA's Jet Propulsion Laboratory (JPL) from 2011
to 2015. Dr. Ryoo received the Ph.D. degree from the University of Texas
at Austin in 2008, and the B.S. degree from the Korea Advanced Institute
of Science and Technology (KAIST) in 2004. His paper on robot-centric
activity recognition at ICRA 2016 won the Best Vision Paper award, and his
HRI 2015 paper was one of the two nominees for its Best Enabling
Technology award. In 2017, he founded a computer vision startup company,
EgoVid Inc., focusing on privacy-preserving recognition of humans,
objects, and their activities.
Prof. Dariu M. Gavrila
received the PhD degree
in computer science from
the University of Maryland
at College Park, USA, in
1996. From 1997 till 2016
he has been with Daimler
R&D in Ulm, Germany, where
he eventually became a
Distinguished Scientist.
In 2010, he was appointed
professor at the University
of Amsterdam, chairing
the area of Intelligent
Perception Systems (part-time).
Since 2016 he heads the
Intelligent Vehicles and
Cognitive Robotics section
at the TU Delft as a Full
Professor. Over the past
15 years, Prof. Gavrila
has focused on visual systems
for detecting humans and
their activity, with application
to intelligent vehicles,
smart surveillance and
social robotics. He led
the multi-year pedestrian
detection research effort
at Daimler, which was incorporated
in the Mercedes-Benz S-,
E-, and C-Class models
(2013-2014). He is frequently
cited in the scientific
literature and received
the I/O 2007 Award from
the Netherlands Organisation
for Scientific Research
(NWO) and the IEEE Intelligent
Transportation Systems
Application Award 2014.
Programm
13:15-13:30Introduction13:30-14:00Invited TalkRobots Anticipating Future Scene
Michael S. Ryoo, Indiana University14:00-15:05Oral SessionAction Anticipation by
Predicting Future Dynamic
Images
Cristian Rodriguez (Australian
National University),
Basura Fernando (Australian
National University),
Hongdong Li (Australian
National University)
Joint Future Semantic
and Instance Segmentation
Prediction
Camille Couprie (Facebook
AI Research), Pauline
Luc (Facebook AI Research),
Jakob Verbeek (INRIA)
Context Graph based Video
Frame Prediction using
Locally Guided Objective
Prateep Bhattacharjee
(Indian Institute of
Technology Madras),
Sukhendu Das (Indian
Institute of Technology
Madras) Predicting Action Tubes
Gurkirt Singh (Oxford
Brookes University),
Suman Saha (Oxford Brookes
University), Fabio Cuzzolin
(Oxford Brookes University)
Forecasting Hands and
Objects in Future Frames
Chenyou Fan (Indiana
University), Jangwon
Lee (Indiana University),
Michael S Ryoo (Indiana
University) 15:05-15:40Poster Session including
Coffee When will you do what?
- Anticipating Temporal
Occurrences of Activities
Yazan Abu Farha (University
of Bonn), Alexander
Richard (University
of Bonn), Juergen Gall
(University of Bonn)
Motion Prediction with
Gaussian Process Dynamical
Models and Trajectory
Optimization
Philipp Kratzer (University
of Stuttgart), Marc
Toussaint (University
of Stuttgart), Jim Mainprice
(University of Stuttgart,
MPI-IS) R2P2: A ReparameteRized
Pushforward Policy for
Diverse, Precise Generative
Path Forecasting
Nicholas Rhinehart (Carnegie
Mellon University),
Kris M. Kitani (Carnegie
Mellon University),
Paul Vernaza (Carnegie
Mellon University) Action Anticipation with
RBF Kernelized Feature
Mapping RNN
Yuge Shi (Australian
National University),
Basura Fernando (Australian
National University),
Richard Hartley (Australian
National University)
Predicting Future Instance
Segmentations by Forecasting
Convolutional Features
Camille Couprie (Facebook
AI Research), Pauline
Luc (Facebook AI Research),
Yann LeCun (Facebook
AI Research), Jakob
Verbeek (INRIA) 15:40-16:20Poster Session including
Coffee Learning to Forecast
and Refine Residual
Motion for Image-to-Video
Generation
Long Zhao (Rutgers University),
Xi Peng (Binghamton
University), Yu Tian
(Rutgers University),
Mubbasir Kapadia (Rutgers
University), Dimitris
Metaxas (Rutgers University)
Deep Video Generation,
Prediction and Completion
of Human Action Sequences
Haoye Cai (Hong Kong
University of Science
and Technology, Stanford
University), Chunyan
Bai (Hong Kong University
of Science and Technology,
Carnegie Mellon University),
Yu-Wing Tai (Tencent
Youtu), Chi-Keung Tang
(Hong Kong University
of Science and Technology)
Adversarial Geometry-Aware
Human Motion Prediction
Liang-Yan Gui (Carnegie
Mellon University),
Yu-Xiong Wang (Carnegie
Mellon University),
Xiaodan Liang (Carnegie
Mellon University),
José M. F. Moura (Carnegie
Mellon University) Embarrassingly Simple
Model for Early Action
Proposal
Marcos Baptista-Ríos
(University of Alcalá),
Roberto Lopez-Sastre
(University of Alcalá),
Francisco J. Acevedo-Rodríguez
(University of Alcalá),
Saturnino Maldonado-Bascon
(University of Alcalá)
Am I done? Predicting
Action Progress in Video
Federico Becattini (University
of Florence), Lorenzo
Seidenari (University
of Florence), Tiberio
Uricchio (University
of Florence), Alberto
Del Bimbo (University
of Florence), Lamberto
Ballan (University of
Padova) A Novel Semantic Framework
for Anticipation of
Manipulation Actions
Fatemeh Ziaeetabar (University
of Göttingen), Minija
Tamosiunaite (University
of Göttingen), Florentin
Wörgötter (University
of Göttingen) 16:20-16:50Invited TalkVulnerable Road User Path
Prediction
Dariu M. Gavrila, Delft
University of Technology
16:50-17:45Oral SessionRED: A simple but effective
Baseline Predictor for
the TrajNet Benchmark
Stefan Becker (Fraunhofer
IOSB), Ronny Hug (Fraunhofer
IOSB), Wolfgang Hübner
(Fraunhofer IOSB), Michael
Arens (Fraunhofer IOSB)
Convolutional Neural
Network for Trajectory
Prediction
Nishant Nikhil (Indian
Institute of Technology
Kharagpur), Brendan
Morris (University of
Nevada Las Vegas) Leader’s Gaze Behaviour
and Alignment of the
Action Planing from
the Follower’s Gaze
Cues in Human-Human
and Human-Robot Interaction
Nuno Duarte (Instituto
Superior Técnico), Mirko
Rakovic (Instituto Superior
Técnico), Jorge Marques
(Instituto Superior
Técnico), José Santos-Victor
(Instituto Superior
Técnico) Group LSTM: Group Trajectory
Prediction in Crowded
Scenarios
Niccolò Bisagno (Università
di Trento), Bo Zhang
(Dalian Maritime University),
Nicola Conci (UNITN)
17:45-18:00Closing Remarks & Discussion
Sergio Escalera University of Barcelona,
Spain
David Fouhey UC Berkeley, United States
of AmericaHildegard Kühne University of Bonn, GermanyThomas B. Moeslund Aalborg University, DenmarkSilvia Pintea Delft University of Technology,
The NetherlandsNick Rhinehart Carnegie Mellon University,
United States of America
Michael S. Ryoo Indiana University Bloomington,
United States of AmericaLeonid Sigal University of British Columbia,
Canada
Gul Varol
INRIA, FranceCarl Vondrick Google, United States of
America
Jiajun Wu
Massachusetts Institute of Technology,
United States of AmericaAngela Yao
University of Bonn, Germany
Contact
Prof. Dr. Jürgen Gall
University of Bonn
Institute of Computer Science
Computer Vision
Group Endenicher Allee 19a
53115 Bonn, Germany
E-mail: ahb2018@lists.iai.uni-bonn.de
Liability note
Despite careful content control
we assume no liability for
the content of external links.
For the corresponding content
of linked pages, their administrators
are responsible.
Prof. Michael S. Ryoo
is an Assistant Professor in the Department of Computer
Science at Indiana University Bloomington. His research interest is within
the areas of Computer Vision and Robotics, with a particular emphasis on
human activity recognition/learning, first-person vision, and robot action
learning. Before joining IU, Dr. Ryoo was a staff researcher at the
Robotics Section of the NASA's Jet Propulsion Laboratory (JPL) from 2011
to 2015. Dr. Ryoo received the Ph.D. degree from the University of Texas
at Austin in 2008, and the B.S. degree from the Korea Advanced Institute
of Science and Technology (KAIST) in 2004. His paper on robot-centric
activity recognition at ICRA 2016 won the Best Vision Paper award, and his
HRI 2015 paper was one of the two nominees for its Best Enabling
Technology award. In 2017, he founded a computer vision startup company,
EgoVid Inc., focusing on privacy-preserving recognition of humans,
objects, and their activities.
Prof. Dariu M. Gavrila
received the PhD degree
in computer science from
the University of Maryland
at College Park, USA, in
1996. From 1997 till 2016
he has been with Daimler
R&D in Ulm, Germany, where
he eventually became a
Distinguished Scientist.
In 2010, he was appointed
professor at the University
of Amsterdam, chairing
the area of Intelligent
Perception Systems (part-time).
Since 2016 he heads the
Intelligent Vehicles and
Cognitive Robotics section
at the TU Delft as a Full
Professor. Over the past
15 years, Prof. Gavrila
has focused on visual systems
for detecting humans and
their activity, with application
to intelligent vehicles,
smart surveillance and
social robotics. He led
the multi-year pedestrian
detection research effort
at Daimler, which was incorporated
in the Mercedes-Benz S-,
E-, and C-Class models
(2013-2014). He is frequently
cited in the scientific
literature and received
the I/O 2007 Award from
the Netherlands Organisation
for Scientific Research
(NWO) and the IEEE Intelligent
Transportation Systems
Application Award 2014.
Prof. Juergen Gall
University of Bonn
Prof. Jan van Gemert
Delft University of Technology
Prof. Kris Kitani
Carnegie Mellon University