Transparent & Reflective objects In the wild Challenges
October 3rd, morning session
Pieter Claesz, Public domain, via Wikimedia Commons
Overview
From industry to the home, vision systems are expected to provide relevant information for every object they may encounter in their environment. Vision-based object understanding in particular has seen significant progress toward this goal as demonstrated, for example, by recent results in challenges such as the COCO, and its follow-up LVIS challenge, or the Benchmark for 6D Object Pose Estimation.
Transparent, reflective and, generally speaking, non-lambertian objects still present major challenges for methods as well as RGB and depth sensors. Color images become dependent on the viewpoint and background. Similarly, readings from depth cameras are unreliable whether they are based on stereo matching, time-of-flight or structured light. The unpredictable and unreliable data is problematic at two different levels. On one hand, such objects will not reliably be understood but their presence in the scene degrades the understanding of other objects in the scene through reflections and light path deformation.
The workshop on Transparent & Reflective objects In the wild Challenges (TRICKY) will discuss object classification, detection, tracking, reconstruction, depth and pose estimation from imaging data for such tricky objects with the aim of advancing the state-of-the-art and fostering novel research directions. A major focus will be put on the applicability of methods in unconstrained scenarios. Check the program for more details.
We welcome virtual participation through our Zoom meeting:
Meeting ID: 692 6165 6200
Password: CCYM11DE
Invited Speakers
Roboception
”Perception challenges in flexible production for the industry perspective”
NVIDIA Robotics Lab
“RGB-D Local Implicit Function for Depth Completion of Transparent Objects”
Carnegie Mellon University
”Self-supervised learning to manipulate transparent and reflective objects.”
Carnegie Mellon University
“Using and Evolving Neural Radiance Fields (NeRF) to Grasping Transparent Objects”
UC San Diego
“3D Reconstruction and Interaction with Complex Light Transport”
Submission
Call for contributed papers
Submission - July 5th July 19th 11:59PM Pacific Time
Author Notification - July 26th 11:59PM Pacific Time
Camera ready - August 2nd 11:59PM Pacific Time
Workshop - October 3rd 2023, morning session
Submission website - EasyChair link
Guidelines
We invite submission of 4 page (excluding references) extended abstracts (following the ICCV 2023 template) on topics related to transparent and reflective object understanding. Reviewing of abstract submissions will be double-blind. The purpose of this workshop is not to be a venue for publication but rather a place to discuss and open new directions of research for transparent and reflective objects understanding, as well as gather a community of people interested in the field. The workshop proceedings will not appear in the official ICCV 2023 workshop proceedings. Submissions of work which has been previously published, including papers accepted to the main ICCV 2023 conference are allowed.
As part of the workshop schedule, 6 submitted papers will be selected for a spotlight presentation as contributed talks, and all accepted abstracts will be presented during the poster session.
Scope
Relevant topics for this workshop include transparent and or reflective objects:
Classification
Detection
Tracking
Pose estimation
Depth estimation
Reconstruction
Sensing
Program
We are happy to announce that the following papers will be presented during the spotlight session of the workshop:
"Polarimetric Information for Multi-Modal 6D Pose Estimation of Photometrically Challenging Objects with Limited Data", Patrick Ruhkamp, Daoyi Gao, Hyunjun Jung, Nassir Navab and Benjamin Busam
"Multi-Modal Dataset Acquisition for Photometrically Challenging Objects", Hyunjun Jung, Patrick Ruhkamp, Nassir Navab and Benjamin Busam
"Learning Residual NeRFs for Transparent Objects in Known Scenes", Bardienus Duisterhof and Jeffrey Ichnowski
"TRansPose: Large-Scale Multispectral Dataset for Transparent Object", Jeongyun Kim, Myung-Hwan Jeon, Sangwoo Jung, Wooseong Yang, Minwoo Jung, Jaeho Shin and Ayoung Kim
"NEMTO: Neural Environment Matting for Novel View and Relighting Synthesis of Transparent Objects", Dongqing Wang, Tong Zhang and Sabine Süsstrunk
"Ref-DVGO: Reflection-Aware Direct Voxel Grid Optimization for an Improved Quality-Efficiency Trade-Off in Reflective Scene Reconstruction", Georgios Kouros, Minye Wu, Shubham Shrivastava, Sushruth Nagesh, Punarjay Chakravarty and Tinne Tuytelaars
Contact
For any inquiry, please contact us at tricky.objects.workshop@gmail.com
Workshop Organizers
Jean-Baptiste Weibel, TU Wien, Vienna, Austria
Doris Antensteiner, Austrian Institute of Technology, Austria
Timothy Patten, University of Technology Sydney, Australia
Jon Azpiazu, TECNALIA Research & Innovation, Donostia, Spain
Markus Vincze, TU Wien, Vienna, Austria