Aus Aifbportal
Wechseln zu:Navigation, Suche

Automated/Autonomous Driving - Aktives Machine Learning für die Objekt Detektion

Information on the Thesis

Type of Final Thesis: Bachelor, Master
Supervisor: Jens Weber
Research Group: Applied Technical Cognitive Systems

Archive Number: 4.756
Status of Thesis: Open
Date of start: 2021-03-31

Further Information

Supervised machine learning algorithms are used in automated driving, the success of those algorithms depends to a large extent on the training data used. To make the training of the ML-algorithms as efficient as possible and to minimize the amount of necessary training data, active learning methods are used. For this, the ML algorithm itself decides which data is relevant for training and which is not. In particular, there is a great need for research in the context of Deep Learning based object detection. In your work, you will develop an active learning system for object detection and evaluate your prototype on a real system in a large-scale infrastructure. You will gain deep insights into Deep Learning based object detection methods as well as in the area of Active Learning.

TASKS You can expect:

  • Exploring the state of research on active learning in the context of object detection
  • Designing an architecture for active learning from an infrastructure perspective
  • Realization of a prototype of the conceptualized architecture
  • Evaluation of your prototype on a real system


  • An interdisciplinary research environment with partners from science and industry
  • A constructive collaboration with bright, motivated employees
  • A pleasant working atmosphere
  • Modern hardware
  • Openness to creative ideas


  • Knowledge of deep learning and computer vision (esp. object detectors such as Yolo, SSD, etc.)
  • Familiarity with Python or C++
  • Familiarity with ROS and Linux advantageous
  • Sound English or German skills
  • High creativity and productivity
  • Knowledge in the field of artificial intelligence (especially searching and learning), game theory or related areas are a plus
  • Experiences with methods for searching and learning, e.g. Monte Carlo tree search/Reinforcement Learning are a plus


  • current transcript of records
  • CV


Jens Weber