The Lexis (Learning EXpert Intelligent Systems) Team depends to the Institute of Measurement, Control and Microtechnology of the Ulm University.
Our research is motivated by problems in machine learning and particularly deep learning, as well as applied machine learning for robotics, autonomous systems, computer vision, medical image analysis and artificial intelligence. Our basic research deals with topics such as representation learning, optimization, uncertainty estimation, multi-modal learning, learning with different forms of supervision (unsupervised, self-supervised), learning algorithm for noisy labels, few-shot learning and meta-learning. Also, we develop optimization algorithms for resource-constrained machine learning.
Our application area includes microscopy image cell segmentation, human behavior understanding (trajectory, body pose and activity) and autonomous driving tasks such as motion estimation and localization, object detection, segmentation and pose estimation, signal denoising and neural architecture search.
Youssef Dawoud, Julia Hornauer, Gustavo Carneiro, Vasileios Belagiannis, Few-Shot Microscopy Image Cell Segmentation, Joint European Conference on Machine Learning and Knowledge Discovery in Databases, (to appear).Pre-print Code Bib File
Wiederer, Julian, Bouazizi, Arij, Kressel, Ulrich, Belagiannis, Vasileios, Traffic Control Gesture Recognition for Autonomous Vehicles, 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), (to appear).
In this work, we address the limitation of the existing autonomous driving datasets to provide learning data for traffic control gesture recognition. We introduce a dataset that is based on 3D body skeleton input to perform traffic control gesture classification on every time step. To evaluate our dataset, we propose eight sequential processing models based on deep neural networks such as recurrent networks, attention mechanism, temporal convolutional networks and graph convolutional networks. We present an extensive evaluation and analysis of all approaches for our dataset, as well as real-world quantitative evaluation. The code and dataset is publicly available.Pre-print Code Data Bib File