GNHK: A Dataset for English Handwriting in the Wild

Abstract

In this paper, we present the GoodNotes Handwriting Kollection (GNHK) dataset. The GNHK dataset includes unconstrained camera-captured images of English handwritten text sourced from different regions around the world. The dataset is modeled after scene text datasets allowing researchers to investigate new localisation and text recognition techniques. We presented benchmark text localisation and recognition results with well-studied frameworks. The dataset and benchmark results are available at https://github.com/GoodNotes/GNHK-dataset.