Campus-Institut Data Science (CIDAS)

Graduation projects

Historical albums with picture postcards

Supervisor: Prof. Langner

Area: Digital Edition

Tools: MySQL and Excel, heuristnetwork, openCV

Task: Make a digital edition of an album and analyse the motifs and locations using a database (and perhaps object recognition).

Lifestyles in chronological comparison: 50 years of IKEA

Supervisor: Prof. Langner

Area: Quantitative Image Type Analysis
Tools: Python, keras, tensorflow, Orange

Task: Reduce the photos in the IKEA catalogue to a few types of images (e.g. family at the dining table, children in the bedroom) and examine their development over the past 50 years.

Acquisition of the SUB collection of historical globes

Supervisor: Prof. Langner

Area: 3D Digitisation

Tools: Metashape, Meshlab, blender


Task: Find suitable methods of 3D-digitisation for these sphaerical objects.

Automatic determination of paper qualities

Supervisor: Prof. Langner

Area: 3D Digitisation

Tools: Structured Light Scanner, Metashape, MeshLab, Python

Task: Develop a workflow to automatically determine the different material qualities of paper (e.g. handmade paper, copperplate paper, etc.) during scanning.

3D shape similarity and resemblance

Supervisor: Prof. Langner

Area: 3D Shape Analysis

Tools: MeshLab, Python


Task: Dimensional comparisons on ancient and post-ancient sculpture based on 3D scans.

Distant Reading and Viewing

Supervisor: Prof. Langner

Area: Multimodality

Tools: Machine Learning (keras, tensorflow), Python, orange, Stanford NLP Tools
Task: Compare mechanisms of story telling using a corpus of Shakespeare illustrations or historical magazines.

Reconstructing ancient lifestyles

Supervisor: Prof. Langner

Area: Social Network Analysis

Tools: gephi, Excel


Task: Use different lists of house inventories from ancient Pompeii (or the Greek city of Olynth) to reconstruct the way of life, social and cultural self-conception and behavior of the inhabitants.

Early Modern Prints

Supervisor: Prof. Wesche

Area: OCR

Tools: Transkribus or OCR4all


Task: Train a model for letter recognition in early modern prints.

Style, motif and meaning in early 20th century art

Supervisor: Prof. Langner

Area: Style Detection

Tools: openCV, Python, keras, tensorflow

Task: Analyse the relationship between styles and motifs in German Expressionism based on 2D image pattern recognition.

VR4Study

Supervisor: Prof. Langner

Area: 3D Modelling

Tools: blender


Task: Create an Avatar for the e-learning application VR4Study.

Historical cycle routes

Supervisor: Prof. Langner

Area: Spatial Analysis

Tools: QGIS


Task: Reconstruction of the bicycle routes from the "Wegweiser durch Mitteldeutschland für Radfahrer" from 1897.

What does good advertising look like?

Supervisor: Prof. Langner

Area: Image Feature Detection
Tools: Fiji (ImageJ) or Python, keras, tensorflow, Orange
Task: Compare formal features of award winning print ads using the Clio-Awards

How much 3D information is contained in depth-of-field photos?

Supervisor: Prof. Langner

Area: 3D-Modelling and 3D Shape Comparison
Tools: Meshlab and Metashape, Python
Task:

Create an "ideal portrait" of Augustus from 20 3D scans of the emperor, map four 2D orthophotos, then create 256 screenshots and calculate a new model photogrammetrically. Then compare the original head with the newly created head using the Hausdorff distance..