Back to all patents
In the public domainaiAssigned to Microsoft Corporation

Method for computer-implemented image understanding using a probabilistic model

Probabilistic graphical model image understanding unlocks royalty-free object and scene inference from video and images.

What it is

The method builds probabilistic graphical models to infer which objects, scenes, or events are present by combining pixel/feature observations with learned relationships. It’s technically significant because it turns visual evidence into explicit probability distributions, enabling structured prediction across many components instead of single-label decisions. Previously, deploying this specific probabilistic-model-based image-understanding workflow could require licensing from the assignee.

What you could build

Build SaaS image- and video-anomaly detection or visual search that uses probabilistic graphical models for structured labeling.

Patent number
7444317B2
Expiration
Invalid Da
Assignee
Microsoft Corporation
Inventors
Christopher Bishop, others

More in ai

Building on this patent?

Submit a problem you're solving with this technology and connect with founders, investors, and researchers.