Improved Oracle Bone Script Detection Using Enhanced SSD and a Novel Dataset

Lin Meng, a researcher at Ritsumeikan University in Japan, has spearheaded the development of a new dataset for advancing oracle bone script detection. This dataset comprises 330 images and addresses the need for robust training data in this specialized field. Recognizing the challenge of accurately detecting smaller oracle bone characters, Lin Meng and the team have implemented enhancements to the Single Shot MultiBox Detector (SSD) algorithm. These improvements aim to significantly boost the accuracy of oracle bone script detection, paving the way for more effective research and analysis of these invaluable historical artifacts.