Theodor Kapler

Papers

Total Citations

H-Index

About

Theodor Kapler is a researcher whose work sits at the intersection of computer vision and probabilistic deep learning, with a particular focus on multi-task learning for autonomous perception. His most cited contribution, "Efficient Multi-task Uncertainties for Joint Semantic Segmentation and Monocular Depth Estimation" (2025), introduces a novel framework that elegantly models task-dependent aleatoric uncertainty to improve the joint performance of two critical vision tasks. By learning to weigh the losses of semantic segmentation and depth estimation dynamically, Kapler’s approach not only boosts accuracy but also provides calibrated uncertainty estimates—a key requirement for safety-critical applications like self-driving cars. Although his publication record is early-stage, with his top paper already garnering 4 citations in its first year, the work demonstrates a clear impact by addressing a fundamental challenge in multi-task learning: balancing competing objectives without manual tuning. Kapler’s research is notable for its practical elegance, offering a computationally efficient solution that can be readily integrated into existing perception pipelines. For students and researchers exploring uncertainty quantification or multi-modal vision, Kapler’s work provides a compelling template for building robust, interpretable models.