Michael André Email and Phone Number

Providing deep learning teams with relevant and high quality annotations by building and maintaining an ETL pipeline.

Vehicle tolling.The complete chain from detection, classification, and tracking is part of the work.The main tasks include vehicle classification and axle detection.

Software developer working with ADAS using machine learning.The team is focusing on detecting and tracking traffic participants.

An investigation of how the technique active learning can limit the amount of annotated data in CNN's.The network's own uncertainty of unseen samples is used for future sample selection.Monte Carlo dropout works as an approximation of the model's epistemic uncertainty.The method is compared against random selection and the entropy in the default predictions.Applied on semantic segmentation.

* Implemented new visualization of some algorithms, written in C++* Made an tool to evaluate internal data, written in Python

Lab assistant in the course Switching Theory and Logical Design at Linköping University.

Mentor for a class in the course Linear Algebra.

Developing a web application to filter a customer database.Is used for creating reports based on specific parameters such as gender, age and municipality.The system is written in SQL, JavaScript and C#.