PyTorch Printer Classification model

1920s Korean Printshop identifier

A simple Django web app with a pretrained PyTorch ResNet50 model will try to classify jpg/png images of book pages from the 1920s Colonial Korea.

The model was trained on a set of ~2200 pages of books with known printers from the 1920s. The model expects an image of a text hailing from Colonial Korea. It aims to classify each image to a printshop, hence uploading images of anything other than a text from this era is pointless.

Currently, it can only differentiate between the printer Hansong Toso (한성도서주식회사 - 漢城圖書株式會社) and Taedong Inswaeso (대동인쇄소 - 大東印刷所).

Further information: GitHub

For more about this project: here