18363798. ELECTRONIC DEVICE AND METHOD WITH POSE PREDICTION simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)
Contents
ELECTRONIC DEVICE AND METHOD WITH POSE PREDICTION
Organization Name
Inventor(s)
ELECTRONIC DEVICE AND METHOD WITH POSE PREDICTION - A simplified explanation of the abstract
This abstract first appeared for US patent application 18363798 titled 'ELECTRONIC DEVICE AND METHOD WITH POSE PREDICTION
Simplified Explanation
An electronic device and method are provided for predicting a pose of an object. The method involves extracting a feature map from an input image, using an ordinal regression model to predict the coordinates of a joint point from the feature map, and then predicting the pose of the object using the predicted coordinates of the joint point.
- The method involves extracting a feature map from an input image.
- An ordinal regression model is used to predict the coordinates of a joint point from the feature map.
- The predicted coordinates of the joint point are then used to predict the pose of the object.
Potential applications of this technology:
- Human pose estimation: This technology can be used to accurately predict the poses of humans in images or videos, which can be useful in various applications such as motion tracking, virtual reality, and gaming.
- Object recognition: By predicting the poses of objects, this technology can aid in object recognition tasks, allowing for more accurate and efficient identification of objects in images or videos.
- Robotics: This technology can be applied in robotics to enable robots to understand and interact with the environment by predicting the poses of objects or humans.
Problems solved by this technology:
- Accurate pose estimation: This technology addresses the challenge of accurately predicting the poses of objects or humans in images or videos, which is a complex task due to variations in appearance, lighting conditions, and occlusions.
- Efficient prediction: The use of a feature map and an ordinal regression model allows for efficient prediction of joint point coordinates and poses, reducing computational complexity compared to other methods.
Benefits of this technology:
- Improved accuracy: By using an ordinal regression model, this technology can provide more accurate predictions of joint point coordinates and poses, leading to better results in applications such as motion tracking or object recognition.
- Real-time performance: The efficient prediction process enables real-time performance, making this technology suitable for applications that require fast and continuous pose estimation, such as gaming or robotics.
- Versatility: This technology can be applied to various objects or humans, making it versatile and adaptable to different scenarios and domains.
Original Abstract Submitted
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.
