Techniques for identifying visually similar content
US10176198B1 · kind B1 · utility
Assignee
Inventors
Key dates
| Filing date | May 9, 2016 |
| Grant date | Jan 8, 2019 |
| Priority date | — |
| Expiry date | Jul 20, 2037 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06V10/764
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
Embodiments of the present invention provide techniques for identifying and recommending similar items based on visual similarity to a selected content item. Visual similarity may be characterized by identifying features depicted in a selected content item and comparing those features to features in an electronic catalog of content items. Visual similarity may also factor in medium and subject matter. For example, when a content item depicts a landscape painting, other landscape paintings (rather than paintings of different subject matter or photographs) will be recommended. Other visual characteristics, such as color theme and distribution, brushwork, etc. may also be represented in the recommended content items. As discussed further herein, different features may be weighted differently based on the analysis of the content item. These weightings enable the recommended content items to be tailored to visually similar subject matter.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.