17987152. METHOD AND APPARATUS WITH FACE LANDMARK COORDINATE PREDICTION simplified abstract (SAMSUNG ELECTRONICS CO., LTD.)
METHOD AND APPARATUS WITH FACE LANDMARK COORDINATE PREDICTION
Organization Name
Inventor(s)
Seon-Min Rhee of Suwon-si (KR)
METHOD AND APPARATUS WITH FACE LANDMARK COORDINATE PREDICTION - A simplified explanation of the abstract
This abstract first appeared for US patent application 17987152 titled 'METHOD AND APPARATUS WITH FACE LANDMARK COORDINATE PREDICTION
Simplified Explanation
The patent application describes a method and apparatus for landmark coordinate prediction in face images using a multi-stage convolutional network and cascaded decoder networks.
- The method involves generating a multi-stage feature map for landmarks of a face image through a staged convolutional network.
- An initial query matrix is generated by fully connecting the last-stage feature map using a fully connected network.
- The total number of feature elements in the initial query matrix is equal to the total number of predicted landmarks of the face image.
- A memory feature matrix is generated by flattening and connecting the multi-stage feature map.
- The predicted landmark coordinates are generated by inputting the memory feature matrix and the initial query matrix to a decoder network of plural cascaded decoder networks.
Potential Applications
- Facial recognition systems
- Facial landmark detection in image processing
- Virtual reality and augmented reality applications
Problems Solved
- Accurate prediction of landmark coordinates in face images
- Efficient feature extraction and processing for facial landmark detection
- Improved performance and accuracy of facial recognition systems
Benefits
- More accurate and reliable facial landmark detection
- Faster and more efficient processing of facial images
- Improved performance of facial recognition systems
Original Abstract Submitted
A method and apparatus with landmark coordinate prediction are provided. The method includes generating a multi-stage feature map for landmarks of a face image through a staged convolutional network, generating an initial query matrix by fully connecting a last-stage feature map in the multi-stage feature map using a fully connected network, where a total number of feature elements in the initial query matrix is equal to a total number of predicted landmarks of the face image, generating a memory feature matrix by flattening and connecting the multi-stage feature map, generating the predicted landmark coordinates by inputting the memory feature matrix and the initial query matrix to a decoder network of plural cascaded decoder networks.
