Unsupervised Structure-Texture Separation Network for Oracle Character Recognition

计算机科学 人工智能 判别式 Glyph(数据可视化) 甲骨文公司 模式识别(心理学) 分类器(UML) 人工神经网络 学习迁移 可视化 软件工程
作者
Mei Wang,Weihong Deng,Cheng‐Lin Liu
出处
期刊:IEEE transactions on image processing [Institute of Electrical and Electronics Engineers]
卷期号:31: 3137-3150 被引量:23
标识
DOI:10.1109/tip.2022.3165989
摘要

Oracle bone script is the earliest-known Chinese writing system of the Shang dynasty and is precious to archeology and philology. However, real-world scanned oracle data are rare and few experts are available for annotation which make the automatic recognition of scanned oracle characters become a challenging task. Therefore, we aim to explore unsupervised domain adaptation to transfer knowledge from handprinted oracle data, which are easy to acquire, to scanned domain. We propose a structure-texture separation network (STSN), which is an end-to-end learning framework for joint disentanglement, transformation, adaptation and recognition. First, STSN disentangles features into structure (glyph) and texture (noise) components by generative models, and then aligns handprinted and scanned data in structure feature space such that the negative influence caused by serious noises can be avoided when adapting. Second, transformation is achieved via swapping the learned textures across domains and a classifier for final classification is trained to predict the labels of the transformed scanned characters. This not only guarantees the absolute separation, but also enhances the discriminative ability of the learned features. Extensive experiments on Oracle-241 dataset show that STSN outperforms other adaptation methods and successfully improves recognition performance on scanned data even when they are contaminated by long burial and careless excavation.

科研通智能强力驱动
Strongly Powered by AbleSci AI
科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
1秒前
斯寜应助TTTTTT采纳,获得10
1秒前
2秒前
Meidina发布了新的文献求助10
2秒前
ll完成签到 ,获得积分10
3秒前
豆子发布了新的文献求助10
5秒前
蓝天白云完成签到,获得积分20
5秒前
瑶啊瑶发布了新的文献求助10
7秒前
隐形曼青应助Cee采纳,获得10
8秒前
wzzznh完成签到 ,获得积分10
9秒前
心悦完成签到 ,获得积分10
9秒前
FashionBoy应助陈泽宇采纳,获得10
10秒前
小二郎应助科研通管家采纳,获得10
13秒前
Hello应助科研通管家采纳,获得10
13秒前
寒江雪应助科研通管家采纳,获得10
13秒前
小马甲应助科研通管家采纳,获得10
13秒前
在水一方应助科研通管家采纳,获得10
13秒前
赘婿应助科研通管家采纳,获得10
13秒前
传奇3应助科研通管家采纳,获得10
14秒前
14秒前
liu完成签到 ,获得积分10
15秒前
所所应助Meidina采纳,获得10
15秒前
斯文败类应助miaowk采纳,获得10
18秒前
19秒前
19秒前
19秒前
19秒前
zhangheng完成签到,获得积分20
20秒前
zhq发布了新的文献求助10
24秒前
花生酱发布了新的文献求助20
24秒前
stevben完成签到,获得积分10
26秒前
陈泽宇发布了新的文献求助10
26秒前
27秒前
小么完成签到 ,获得积分10
28秒前
阿赫完成签到 ,获得积分10
29秒前
houfei发布了新的文献求助10
29秒前
苗条的水儿完成签到,获得积分10
30秒前
xiaoguanyan完成签到,获得积分10
31秒前
32秒前
miaowk发布了新的文献求助10
32秒前
高分求助中
Chinesen in Europa – Europäer in China: Journalisten, Spione, Studenten 500
Arthur Ewert: A Life for the Comintern 500
China's Relations With Japan 1945-83: The Role of Liao Chengzhi // Kurt Werner Radtke 500
Two Years in Peking 1965-1966: Book 1: Living and Teaching in Mao's China // Reginald Hunt 500
Epigenetic Drug Discovery 500
Hardness Tests and Hardness Number Conversions 300
Knowledge management in the fashion industry 300
热门求助领域 (近24小时)
化学 材料科学 医学 生物 工程类 有机化学 物理 生物化学 纳米技术 计算机科学 化学工程 内科学 复合材料 物理化学 电极 遗传学 量子力学 基因 冶金 催化作用
热门帖子
关注 科研通微信公众号,转发送积分 3816948
求助须知:如何正确求助?哪些是违规求助? 3360399
关于积分的说明 10407721
捐赠科研通 3078337
什么是DOI,文献DOI怎么找? 1690720
邀请新用户注册赠送积分活动 814023
科研通“疑难数据库(出版商)”最低求助积分说明 767985