Title: IoT-based dataset augmentation method for oracle bone inscriptions reorganisation

Authors: Tomoki Morioka; Lin Meng

Addresses: Graduate School of Science and Engineering, Ritsumeikan University, 1-1-1 Noji-higashi, Kusatsu Shiga 525-8577, Japan ' College of Science and Engineering, Ritsumeikan University, 1-1-1 Noji-higashi, Kusatsu Shiga 525-8577, Japan

Abstract: Deep learning techniques are wildly used in various areas. Currently, deep learning is attended in cultural heritage preservation, such as oracle bone inscriptions re-organising. Usually, deep learning techniques need rich and pure data for the training model. However, oracle bone inscriptions are weak in their limited resource. Hence, the dataset of oracle bone inscriptions should be augmented, when some new oracle bones are found. To overcome this problem, this paper proposes an IoT-based dataset augmentation method for oracle bone inscription reorganisation. We design an Android application for data collection and a cloud server application for dataset augmentation. Furthermore, an AI-Filter is designed and implemented in the Android application to keep the collected data pure. Experimental results show the oracle bone inscriptions recognition recall increases by 0.15 with the IoT-based dataset augmentation method, which proved the effectiveness of our proposal.

Keywords: dataset augmentation; oracle bone inscriptions; internet of things; IoT.

DOI: 10.1504/IJAMECHS.2023.131338

International Journal of Advanced Mechatronic Systems, 2023 Vol.10 No.2, pp.102 - 111

Received: 09 Jul 2022
Accepted: 19 Feb 2023
Published online: 06 Jun 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article