FY27 Master Thesis Project- Foundation models for pretraining ADAS perception, 6 months, Linköping - Sweden
Core
Investigate and evaluate foundation models and self-supervised pretraining strategies for autonomous driving perception systems, specifically focusing on knowledge distillation for ADAS and label-efficient 3D object detection.
Role type
Master thesis researcher (AI/Computer Vision)
Builds
Experimental evaluation of pretraining strategies and model architectures for ADAS perception
Domain
Autonomous driving, ADAS, Computer Vision
Deliverable
research
Required skills
Machine Learning, Deep Learning, Computer Vision, Python, PyTorch
Responsibilities
Investigate knowledge distillation methods from foundation models to compact ADAS perception models; Evaluate self-supervised pretraining strategies for BEV representations from LiDAR and camera data; Compare model architectures and pretraining approaches across different perception tasks.
