发明名称 |
Unsupervised adaptation method and automatic image classification method applying the same |
摘要 |
An automatic image classification method applying an unsupervised adaptation method is provided. A plurality of non-manually-labeled observation data are grouped into a plurality of groups. A respectively hypothesis label is set to each of the groups according to a classifier. It is determined whether each member of the observation data in each of the groups is suitable for adjusting the classifier according to the hypothesis label, and the non-manually-labeled observation data which are determined as being suitable for adjusting the classifier are set as a plurality of adaptation data. The classifier is updated according to the hypothesis label and the adaptation data. The observation data are classified according to the updated classifier. |
申请公布号 |
US9299008(B2) |
申请公布日期 |
2016.03.29 |
申请号 |
US201314025391 |
申请日期 |
2013.09.12 |
申请人 |
INDUSTRIAL TECHNOLOGY RESEARCH INSTITUTE |
发明人 |
Huang Tai-Hui;Shih Ming-Yu |
分类号 |
G06K9/62 |
主分类号 |
G06K9/62 |
代理机构 |
Muncy, Geissler, Olds & Lowe, P.C. |
代理人 |
Muncy, Geissler, Olds & Lowe, P.C. |
主权项 |
1. An unsupervised adaptation method applied to an electronic device, comprising:
implementing a processor to perform the following steps; grouping a plurality of non-manually-labeled observation data into a plurality of groups; setting a respective hypothesis label to each of the groups according to a classifier; obtaining a shortest distance between each member of the observation data of each of the groups and a plurality of the members of the sample set of the hypothesis label, and obtaining a minimum of a plurality of distances between the member of the observation data and a plurality of members of representation sets of other hypothesis labels; determining whether each member of the observation data in the groups is suitable for adjusting the classifier according to a ratio of a shortest distance between the observation data and the hypothesis label to a minimum of a plurality of distances between the observation data and other hypothesis labels, and setting the observation data which are determined as being suitable for adjusting the classifier as a plurality of adaptation data; predicting at least one adjustment parameter of the classifier according to the hypothesis labels and the adaptation data, to adjust the classifier; and iterating the above steps to adjust the classifier. |
地址 |
Chutung, Hsinchu TW |