Master Thesis
Status: open
Supervisor: Florian Kleber
The aim of this master’s thesis is to develop/analyze and or enhance automated tools for the enrichment of audiovisual (AV) materials (historical films), including metadata extraction and semantic analysis.
The focus is on
- OverScan detection
- Shot boundary detection
- Shot type classification
- Camera movement classification
- Object Detection & tracking
- Text recognition and analysis
- Relation detection
- Subtitling and translation
For further details see also the project description of AVCloud. Within the master thesis, one or a combination of the above topics should be chosen and further developed.
The research consists of
- Literature Review – getting to know the methods
- Implementation & Evaluation
- Evaluate state-of-the-art methods on the provided datasets
- Develop and apply your processing pipeline
- Comparison and thorough evaluation
- Written Thesis and final presentation
- Summarize your work in a publication (optionally).
References
- Tingyu Lin, Christian Stippel, Armin Dadras, Jakob Zenzmaier, Florian Kleber, Wolfgang Aigner, Robert Sablatnig:
PERSIST: Persistent-State Discrimination for Shot Boundary Detection. CoRR abs/2608.29287 (20 - Rafael Sterzinger
, Tingyu Lin, Robert Sablatnig:
Few-Shot Connectivity-Aware Text Line Segmentation in Historical Documents. ACPR (1) 2025: 100-114 - Tingyu Lin, Marco Peer, Florian Kleber, Robert Sablatnig:
ClapperText: A Benchmark for Text Recognition in Low-Resource Archival Documents. ICDAR (Workshops 2) 2025: 329-346 - Tingyu Lin, Armin Dadras, Florian Kleber, Robert Sablatnig:
DGME-T: Directional Grid Motion Encoding for Transformer-Based Historical Camera Movement Classification. SUMAC@ ACM Multimedia 2025: 13-21.
Helpful experience
- Python
- Good understanding of deep learning
- Machine Learning frameworks (PyTorch)