Museums are having a hard time digitalizing their catalogues. They have to manually create a description for the artworks. Object Detection provides one way of relieving this burden and attempts to provide useful input in this field (annotating objects in the artworks).
Goal:
– Provide the current landscape of Object Detection techniques used in this field (what models are used, how do they perform, …)
– Provide a starting point for future research: what direction should the research field take to provide the most value (either for the end user and the researcher; data augmentation?)
Master thesis outline:
– List of images: references to all images used in the thesis
– List of tables references to all tables used in the thesis
– Introduction: describe the challenge and Object Detection (if you want I can write this myself)
– Literature study: describe all the information needed to understand the practicals of the thesis in detail: the different models
– Related works: Saint George on a Bike project (https://www.zooniverse.org/projects/artem-dot-reshetnikov/saint-george-on-a-bike/about/research), Insight project with Minerva dataset (https://hosting.uantwerpen.be/insight/index.php/category/uncategorised/), others?
– Experiments: run different object detection models to see which one performs best based on the metrics. Provide a theoretical approach as to what could improve the results, experiment with techniques (data augmentation).
– Conclusion: summarize the experiments (what object detection approach is best) and how the results can be improved (data augmentation). Discuss techniques that could be used to improve the results based on the experiments.
– References: in IEEE style. You should use 20-50 references.