Tutorials (tentative)

Tutorials will be held at Noyori Conference Hall on May 11.

Tentative Schedule
10:00‒12:00Mohamed OmranLarge-Scale Datasets and Scene Understanding
14:00‒16:00Yukiyasu DomaeMachine Vision for Problems with Robot Manipulation

Large-Scale Datasets and Scene Understanding

Mohamed Omran
Mohamed Omran
Max Planck Institute for Informatics

I will give an overview of large-scale datasets in the last years, and in particular talk about our own experience putting together the Cityscapes dataset and the challenges involved. During the 2nd part of the tutorial I will discuss recent methods for semantic pixel-level and instance-level labelling followed by some practical tips and tricks based on Cityscapes results.

Machine Vision for Problems with Robot Manipulation

Yukiyasu Domae
Yukiyasu Domae
Mitsubishi Electric Corporation Advanced Technology R&D Center

There are many manipulation problems in the field of factory and warehouse automation. We applied various machine vision techniques to the system in order to tackle those challenges. Last year, we competed in the Amazon Picking Challenge with the system we developed. In this tutorial, we will explain the machine vision system and algorithm.

Talk Outline
  1. 10 min: Trends on Factory Automation (FA)
  2. 20 min: Case study: Cell production robot systems
  3. 20 min: Machine vision algorithms for FA
  4. 10 min: Break
  5. 10 min: Trends on Warehouse Automation (WA)
  6. 20 min: Case study: Picking robot systems
  7. 10 min: Machine vision algorithm for WA
  8. 10 min: Other applications
  9. 10 min: Questions