Fully Convolutional Networks for Semantic Segmentation

TL;DR

This paper introduces fully convolutional networks that perform end-to-end pixel-wise semantic segmentation, achieving state-of-the-art results efficiently by adapting classification models and combining multi-layer features.

Contribution

It presents the concept of fully convolutional networks for dense prediction, adapting existing classification architectures, and introduces a novel multi-layer feature integration for improved segmentation accuracy.

Findings

Achieved 62.2% mean IU on PASCAL VOC 2012

Reduced inference time to one third of a second per image

Outperformed previous state-of-the-art methods in semantic segmentation

Abstract

Convolutional networks are powerful visual models that yield hierarchies of features. We show that convolutional networks by themselves, trained end-to-end, pixels-to-pixels, exceed the state-of-the-art in semantic segmentation. Our key insight is to build "fully convolutional" networks that take input of arbitrary size and produce correspondingly-sized output with efficient inference and learning. We define and detail the space of fully convolutional networks, explain their application to spatially dense prediction tasks, and draw connections to prior models. We adapt contemporary classification networks (AlexNet, the VGG net, and GoogLeNet) into fully convolutional networks and transfer their learned representations by fine-tuning to the segmentation task. We then define a novel architecture that combines semantic information from a deep, coarse layer with appearance information from a shallow, fine layer to produce accurate and detailed segmentations. Our fully convolutional network achieves state-of-the-art segmentation of PASCAL VOC (20% relative improvement to 62.2% mean IU on 2012), NYUDv2, and SIFT Flow, while inference takes one third of a second for a typical image.