Systems and methods for image segmentation
US12417646B2 · kind B2 · utility
Assignee
Inventors
Key dates
| Filing date | Aug 16, 2024 |
| Grant date | Sep 16, 2025 |
| Priority date | — |
| Expiry date | Aug 16, 2044 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06V2201/03
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
Provided herein are methods for image segmentation. An image having a plurality of pixels arranged in a plurality of dimensions is obtained. The image is provided to a pretrained machine learning model and a plurality of amplitudes for each of the plurality of pixel is received from the model. Each of the plurality of amplitudes for each of the plurality of pixels corresponds to exactly one of the plurality of dimensions. At least one instance in the image is identified by: determining a first basin associated with the at least one instance, where the basin consists of a contiguous subset of the plurality of pixels having an amplitude with a magnitude not exceeding first threshold, determining a second subset of the plurality of pixels having an amplitude with a magnitude exceeding the first threshold, where the second subset of pixels bounds the first subset of pixels, and constructing the at least one instance by a flow of the second subset of pixels to the first subset of pixels according to the amplitudes thereof.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.