Decoupled Deep Neural Network for Semi-supervised Semantic Segmentation

TL;DR

This paper introduces a decoupled deep neural network architecture for semi-supervised semantic segmentation that separates classification and segmentation tasks, enabling effective learning from heterogeneous annotations and outperforming existing methods.

Contribution

The novel architecture decouples classification and segmentation, allowing separate training and improved performance with less annotated data.

Findings

Outperforms existing semi-supervised methods on PASCAL VOC dataset.

Effectively utilizes class-specific activation maps to reduce segmentation search space.

Achieves high accuracy with fewer strongly annotated images.

Abstract

We propose a novel deep neural network architecture for semi-supervised semantic segmentation using heterogeneous annotations. Contrary to existing approaches posing semantic segmentation as a single task of region-based classification, our algorithm decouples classification and segmentation, and learns a separate network for each task. In this architecture, labels associated with an image are identified by classification network, and binary segmentation is subsequently performed for each identified label in segmentation network. The decoupled architecture enables us to learn classification and segmentation networks separately based on the training data with image-level and pixel-wise class labels, respectively. It facilitates to reduce search space for segmentation effectively by exploiting class-specific activation maps obtained from bridging layers. Our algorithm shows outstanding performance compared to other semi-supervised approaches even with much less training images with strong annotations in PASCAL VOC dataset.