DepTH: Deep Training on History

Dr Seyran Khademi was a researcher-in-residence at the KB in 2020.

His project ‘DepTH: Deep Training on History’ explored the question: How do we simultaneously examine modern and ancient images?

Project

Seyran Khademi explored the domain of illustrations for the broad task of computer vision. Her work resulted in the Ot & Sien Dataset, a dataset of children’s book illustrations from before 1880, manually labelled for the task of object recognition. 

Her blogpost described the process of creating the dataset, as well as the challenges and motivations behind open-set object detection. It additionally links to a baseline classification model based on the dataset.