One-Shot Learning for Semantic Segmentation
This work addresses the challenge of learning semantic segmentation from sparse annotated data, which is incremental as it adapts existing low-shot methods to a new task.
The paper tackles the problem of one-shot semantic segmentation by extending low-shot learning techniques to dense pixel-level prediction, achieving a 25% relative meanIoU improvement and at least 3 times faster performance compared to baselines on the PASCAL VOC 2012 dataset.
Low-shot learning methods for image classification support learning from sparse data. We extend these techniques to support dense semantic image segmentation. Specifically, we train a network that, given a small set of annotated images, produces parameters for a Fully Convolutional Network (FCN). We use this FCN to perform dense pixel-level prediction on a test image for the new semantic class. Our architecture shows a 25% relative meanIoU improvement compared to the best baseline methods for one-shot segmentation on unseen classes in the PASCAL VOC 2012 dataset and is at least 3 times faster.