Curriculum Vitae

Julian Tatsch holds two masters degrees with distinction in computer science (2015) and business (2011) from Technische Universität München. From 12/2015 until 05/2019 he was a research assistant in the computer vision group at the BMW Group Autonomous Driving Department.

 

Research Interests

Scene Understanding with Computer Vision:

  • Improving Semantic Segmentation with Depth/Temporal Cues
  • Semantic Segmentation of Point Clouds
  • Entity Relationship Detection
  • Learning from Artificial Data and Transferring to the Real World
  • Active Learning

 

Activities

Organizer of 1st International Workshop on "Data Driven Intelligent Vehicle Applications" (DDIVA 2019) @ 30th IEEE Intelligent Vehicles Symposium (IV 2019)

Organizer of 2st International Workshop on "Data Driven Intelligent Vehicle Applications" (DDIVA 2020) @ 31th IEEE Intelligent Vehicles Symposium (IV 2020)

Supervised Theses

  • The Value of Depth for Semantic Segmentation Neural Networks
  • Temporal Modeling for Semantic Video Segmentation
  • Point Cloud Segmentation for Automotive Applications
  • Fusing Visual Features and Semantic Knowledge for Visual Relationship Understanding in Autonomous Driving
  • Deep Learning of Knowledge Embeddings for Visual Scene Understanding in Autonomous Driving
  • Active Learning - Intelligent Training Strategies for Data-Efficient Object Detectors in Autonomous Driving
  • Learning from Synthetic Data: Domain Transfer for Detection- and Segmentationmodels

Publications

2020

  • Julian Thomas, Julian Tatsch, Alois Knoll and Raúl Rojas: Online Road Model Generation From Evidential Semantic Grids. 2020 IEEE Intelligent Transportation Systems Conference (ITSC), 2020Rhodes, Greece. September 20-23, 2020 (Virtual), 110-117 mehr… BibTeX Volltext (mediaTUM)
  • Sebstian Schmidt, Julian Tatsch, Qing Rao, Alois Knoll: Advanced Active Learning Strategies for Object Detection. 2020 IEEE Intelligent Vehicles Symposium (IV), 2020June 23-26, 2020, Las Vegas, NV, USA mehr… BibTeX Volltext (mediaTUM)

2019

  • Julian Thomas, Julian Tatsch, Wim van Ekeren, Raúl Rojas, Alois Knoll: Semantic Grid-Based Road Model Estimation for Autonomous Driving. IEEE Intelligent Vehicles Symposium 2019, 2019 mehr… BibTeX Volltext (mediaTUM)
  • Oliver Wasenmüller, René Schuster, Didier Stricker, Karl Leiss, Jürger Pfister, Oleksandra Ganus, Julian Tatsch, Artem Savkin, Nikolas Brasch: Automated Scene Flow Data Generation for Training and Verification. ACM Computer Science in Cars Symposium, 2018, Munich, Germany, 2019 mehr… BibTeX Volltext (mediaTUM)