Pascal Loth
Papers
1
Total Citations
9
H-Index
1
About
Pascal Loth is a researcher at the intersection of computer vision and evolutionary computation, with a primary focus on designing efficient neural architectures for real-world object detection. His most influential work, "Designing Convolutional Neural Networks Using a Genetic Approach for Ball Detection" (2019, 9 citations), introduces a novel methodology that leverages genetic algorithms to automatically evolve convolutional neural network (CNN) architectures tailored for ball detection in sports scenarios. This contribution addresses a critical challenge in automated sports analytics: the need for lightweight, high-performance models that can operate in dynamic environments. By combining the exploratory power of evolutionary search with the representational capacity of deep learning, Loth demonstrates how automated architecture design can outperform manually crafted networks in specific detection tasks. His work has implications for robotics, surveillance, and interactive systems where rapid, accurate object localization is essential. Though early in his career, Loth's research signals a promising direction for integrating neuroevolution with practical computer vision applications, offering a blueprint for researchers seeking to automate model design in resource-constrained settings.
Research Focus
Key Achievements
Top Papers
- 1