System and method for unsupervised learning of segmentation tasks
AssigneeSamsung Electronics Co., Ltd.
InventorsQingfeng Liu, Mostafa El-Khamy, Yuewei Yang
Life of this patent
Estimating the term…
Abstract
Apparatuses and methods are provided for training a feature extraction model determining a loss function for use in unsupervised image segmentation. A method includes determining a clustering loss from an image; determining a weakly supervised contrastive loss of the image using cluster pseudo labels based on the clustering loss; and determining the loss function based on the clustering loss and the weakly supervised contrastive loss.
Claims16
A method of training a feature extraction model by determining a loss function for use in unsupervised image segmentation, the method comprising:
- determining a clustering loss from an image;
- determining a weakly supervised contrastive loss of the image using cluster pseudo labels based on the clustering loss;
- determining a box sample loss of the image;
- and
- determining the loss function based on the clustering loss, the box sample loss, and the weakly supervised contrastive loss.
An apparatus for training a feature extraction model by determining a loss function for use in unsupervised image segmentation, the apparatus comprising:
- a processor;
- and
- a memory configured to store instructions, which when executed, control the processor to:
- determine a clustering loss from an image,
- determine a weakly supervised contrastive loss of the image using cluster pseudo labels based on the clustering loss,
- determine a box sample loss of the image, and
- determine the loss function based on the clustering loss, the box sample loss, and the weakly supervised contrastive loss.
A method of training a feature extraction model by determining a loss function for use in unsupervised image segmentation, the method comprising:
- determining a clustering loss from an image;
- determining a box sample loss of the image;
- and
- determining the loss function based on the clustering loss and the box sample loss.
An apparatus for training a feature extraction model by determining a loss function for use in unsupervised image segmentation, the apparatus comprising:
- a processor;
- and
- a memory configured to store instructions, which when executed, control the processor to:
- determine a clustering loss from an image,
- determine a box sample loss of the image, and
- determine the loss function based on the clustering loss and the box sample loss.
Dependencies are inferred from the claim text.