Master's Thesis: Human-in-the-Loop Deep Learning for Radar-Based Object Dimensioning

Responsibilities

• Develop a Deep Learning pipeline for object dimensioning based on radar data and human feedback.
• Investigate methods to integrate radar information with human feedback for efficient model training.
• Compare innovative labeling approaches, such as Pairwise Preference Learning and directional feedback.
• Implement an interactive UI tool to facilitate efficient annotation by human experts.
• Evaluate developed methods regarding accuracy, robustness, and annotation effort.

Requirements

• Currently enrolled student in Computer Science, Electrical Engineering, Robotics, Data Science, or a comparable field.
• Proficiency in Python programming.
• Practical experience with Deep Learning frameworks, preferably PyTorch or JAX.
• Basic understanding of Computer Vision (e.g., camera projections, coordinate systems) or radar signal processing.
• Enthusiastic, team-oriented, and able to work independently.

About the Company

FORVIA HELLA is an internationally active automotive supplier and part of the FORVIA Group. We specialize in high-performance lighting technology and vehicle electronics, providing a wide range of products and services for the automotive industry, including automated driving solutions.