Franz Köferl, M. Sc.

Department of Computer Science
Lehrstuhl für Informatik 14 (Maschinelles Lernen und Datenanalytik)

Room: Room 01.015
Carl-Thiersch-Straße 2b
91052 Erlangen

Since 7/2017 PhD Candidate

Machine Learning and Data Analytics Lab,
Department of Computer Science, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU)

5/2014 – 2/2017 M.Sc. in Computer Science
Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU)
Master Thesis : “Segmentation and Classification of Interlinear and Marginal Glosses using Convolutional Neural Networks
7/2016 – 6/2017 Student/Assistant Researcher at Fraunhofer IIS
5/2011 – 5/2014
B.Sc. in Computer Science
Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU)

Bachelor Thesis: “Experimentelle Untersuchung des Einflusses von Bound-Handling-Strategien auf die Partikel-Verteilung bei der Partikelschwarmoptimierung.”

Demonstrator Deep Learning

Machine Learning, especially Deep Learning are complex methods for solving various problems, ranging from detection of persons to predicting the time for maintenance of specific parts of a machine. The methods struggle with intuition, when you can solve or when you can’t apply these methods to solve a specific task. This demo demonstrates the conditions needed for a reliable application of deep learning methods and their respective results. Have look at our poster.

The following videos demonstrate an Object Detector Yolov3 [1] trained on naive data, augmentated and simulated from our demo machine. The used data consists of ca. 40,000 images, each manually labeled.

  1. Trained without any measure
  2. Trained with patch augmentation
  3. Trained with simulated images

[1] Joseph RedmonAli Farhadi: YOLOv3: An Incremental Improvement.

Partners & Funding Agencies