The workshop will be held on 30th September 2019 at the Technical School in Zadar, Croatia. The technical program will take place in room 201.
| 16:30 | Opening |
| 16:35 |
Introduction to deep learning for generative modeling (i)
Jakob Verbeek, INRIA Grenoble Abstract: Visual recognition has experienced drastic transformationss due to the adoption of machine learning techniques since the early 2000's, and in particular the widespread adoptions of deep Convolutional Neural Networks (CNNs) since 2012. State of the art approaches for tasks such as object detection, image retrieval, or semantic segmentation are now invariably based on CNNs. While extremely successful, supervised deep learning techniques such as CNNs have a number of limitations, including the requirement of large labelled training data sets. In this tutorial we focus on unsupervised deep learning techniques for visual data. These are of interest for a number of reasons, including (i) to learn visual representations without the need for supervised training data, (ii) to learn generative models to synthesize new images, (iii) as a tool to allow structured prediction in supervised tasks. This tutorial covers the basic principles of deep learning and gives an overview of the main paradigms in unsupervised deep learning, including generative adversarial networks, variational auto-encoders, autoregressive models, and invertible flow based models. |
| 17:35 | Short break |
| 17:45 |
Introduction to deep learning for generative modeling (ii)
Jakob Verbeek, INRIA Grenoble |
| 18:45 | Short break |
| 18:55 |
Simultaneous Semantic Segmentation and Outlier Detection
in Presence of Domain Shift
Petra Bevandić, UniZg-FER Originally presented at GCPR 2019 (pdf). |
| 19:00 |
In Defense of Pre-trained ImageNet Architectures
for Real-time Semantic Segmentation of Road-driving Images
Marin Oršić, UniZg-FER Originally presented at CVPR 2019 (pdf). |
| 19:05 |
Single Level Feature-to-Feature Forecasting
with Deformable Convolutions
Josip Šarić, UniZg-FER Originally presented at GCPR 2019. (pdf). |
| 19:10 |
Efficient Ladder-style DenseNets
for Semantic Segmentation of Large Images
Lucija Ivković, UniZg-FER Partially published on Arxiv (pdf). |
| 19:15 | Discussion |
| 19.30 | Closing |
This workshop seizes an opportunity of Jakob Verbeek visiting Zadar to participate at the Breaking the Surface Workshop in Biograd. The workshop offers an opportunity for the exchange of recent work between INRIA Grenoble and Croatian institutions as well as for fostering future collaboration.