Deep Learning for Oracle Bone Inscriptions Detection: A Supervised Approach

Currently, the detection of oracle bone inscriptions leverages the power of deep learning through a supervised approach. This method involves training deep neural networks using a substantial dataset containing precisely annotated positions of oracle bone inscriptions within images. The training process can be implemented in two primary ways: a one-stage approach or a two-stage approach. Both methods ultimately aim to achieve automatic annotation of oracle bone inscriptions, enabling researchers to analyze and understand these ancient scripts more efficiently.