Fig 2 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Region duplication pretext task for self-supervised learning and feature extraction. The figure illustrates the six different ways of duplicating the regions for a single image patch of 512 x 512: diagonal, off-diagonal, bottom horizontal, top-horizontal, left-vertical and right-vertical.
Source publication
Breast cancer is one of the leading causes of female mortality in the world. This can be reduced when diagnoses are performed at the early stages of progression. Further, the efficiency of the process can be significantly improved with computer aided diagnosis. Deep learning based approaches have been successfully applied to achieve this. One of th...
Context in source publication
Context 1
... generate the "free" labels from the unlabelled data, the 512 x 512 image patch was divided into a 2 x 2 grid. Consider the image patch shown in Figure 2 (top left). In this 2 x 2 grid, there are 6 ways of duplication possible: diagonal, offdiagonal, bottom-horizontal, top-horizontal, left-vertical and right-vertical. ...
Similar publications
Semi supervised learning (SSL) provides an effective means of leveraging unlabelled data to improve a model's performance. Even though the domain has received a considerable amount of attention in the past years, most methods present the common drawback of being unsafe. By safeness we mean the quality of not degrading a fully supervised model when...
