Crowd-OBIGA: A Crowdsourced Approach for Oracle Bone Inscriptions Glyph Annotation

High-quality glyph key point annotation data plays a crucial role in the study of Oracle Bone Inscriptions (OBI) detection and recognition. However, the presence of noise, such as vague scratches and incomplete or irrelevant natural marks on most oracle bone fragments significantly impacts the accuracy of machine algorithms in glyph recognition. Although OBI experts can rely on their professional experience to avoid the impact of noise, the number of experts is small and costly especially for large-scale labeling needs. Crowdsourcing, as a prevalent and cost-effective solution for data annotation leveraging human visual recognition abilities via sophisticated task design, reduces the difficulty and cost of oracle bone glyph annotation. In this paper, we introduce a crowdsourcing tool for annotating glyph key points in OBI. Through a simple interactive interface, crowdsourcing workers can provide key point annotation data, thereby offering abundant training data for machine learning-based oracle bone glyph recognition and classification.