Jan Luca Fahrendholz-Heiermann

Computer Vision and Data Science in the field of construction progress monitoring

Title: Automated progress monitoring on construction site using deep learning based on point clouds and building information models Grade: 1,0 Description: In my master thesis I developed and implemented a fully automated pipeline for progress monitoring on the construction site starting from the planning phase of a building up to active monitoring during construction. To implement and optimize the monitoring, I applied Deep Learning based on 3D point clouds and building information models, among others.

Computer Vision and Data Science in the field of construction progress monitoring

Title: „Machine Vision: Object detection of characteristic features in Tailored Textiles with the help of neural networks“ Grade: 1,7 Description: In my bachelor thesis I developed and implemented an automated optical quality control of tailored textiles. For this purpose, I implemented an object detection on images of tailored textiles with the help of a neural network, whereby defects as well as features in the textile can be detected and displayed.

- Construction and drawing derivation of individual parts and assemblies for mechanical production and prototyping using Catia V5 - Planning and implementation of test setups as well as test execution and analysis - Execution of force/displacement measurements as well as static pull-off force determinations of welded components - Documentation of technical tests from the conception to the analysis of results - Qualitative component analysis and product testing for shipping release

With regard to this master‘s program, I have specialized in topics such as Data Science, Automation as well as Robotic. Among other things, I expanded my knowledge in the area of machine learning techniques using various application examples. This all led to the realization of my master thesis, in which I developed and implemented a complete pipeline for automated progress monitoring on construction sites using Deep Learning based on 3D point clouds and BIM models from real data.