Built-in Foreground/Background Prior for Weakly-Supervised Semantic Segmentation

TL;DR

This paper introduces a novel method for weakly-supervised semantic segmentation that extracts accurate foreground/background masks directly from pre-trained CNN activations, improving localization without external objectness modules.

Contribution

It proposes a new approach to generate masks from CNN activations smoothed by dense CRF, eliminating the need for additional objectness training and enhancing segmentation accuracy.

Findings

Outperforms state-of-the-art tag-based weakly-supervised methods

Uses higher-level CNN activations and dense CRF for mask extraction

Introduces a new inexpensive weak supervision technique

Abstract

Pixel-level annotations are expensive and time consuming to obtain. Hence, weak supervision using only image tags could have a significant impact in semantic segmentation. Recently, CNN-based methods have proposed to fine-tune pre-trained networks using image tags. Without additional information, this leads to poor localization accuracy. This problem, however, was alleviated by making use of objectness priors to generate foreground/background masks. Unfortunately these priors either require training pixel-level annotations/bounding boxes, or still yield inaccurate object boundaries. Here, we propose a novel method to extract markedly more accurate masks from the pre-trained network itself, forgoing external objectness modules. This is accomplished using the activations of the higher-level convolutional layers, smoothed by a dense CRF. We demonstrate that our method, based on these masks and a weakly-supervised loss, outperforms the state-of-the-art tag-based weakly-supervised semantic segmentation techniques. Furthermore, we introduce a new form of inexpensive weak supervision yielding an additional accuracy boost.