Back to results
US 12,700,122 B2Granted
Electronic device and method with pose prediction
AssigneeSamsung Electronics Co., Ltd.
InventorsZhihua Liu, Paul Oh, Qiang Wang, Yamin Mao
Life of this patent
Estimating the term…
Abstract
An electronic device for predicting a pose and a method of operating the electronic device are provided. The method includes extracting a feature map from an input image, predicting coordinates of a joint point from the feature map using an ordinal regression model, and predicting a pose of an object using the predicted coordinates of the joint point.
Claims19
A method of operating an electronic device, the method comprising:
- extracting a feature map from an input image;
- predicting coordinates of a joint point from the feature map using an ordinal regression model, the joint point corresponding to a point of a joint of an object;
- and
- predicting a pose of the object using the predicted coordinates of the joint point,
- wherein the extracting of the feature map comprises:
- inputting the input image to a transformer, and
- extracting from the input image, by the transformer, a first feature map that is an intermediate output of the transformer and a second feature map that is a final output of the transformer.
4 more elements
A method of operating an electronic device, the method comprising:
- extracting a feature map from an input image using a transformer;
- generating, by an ordinal regression model, from the feature map, classification feature volumes, wherein the classification feature volumes are an intermediate output of the ordinal regression model;
- predicting, based on the classification feature volumes, coordinates of a joint point, wherein the predicted coordinates are an output of the ordinal regression model;
- and
- predicting a pose of an object using the predicted coordinates of the joint point.
2 more elements
An electronic device comprising:
- a processor configured to extract a feature map from an input image, predict coordinates of a joint point from the feature map using an ordinal regression model, and predict a pose of an object using the predicted coordinates of the joint point.
Dependencies are inferred from the claim text.