Efficient Dense Modules of Asymmetric Convolution for Real-Time Semantic Segmentation

TL;DR

This paper introduces EDANet, a novel convolutional network that achieves high-speed, low-cost semantic segmentation with accuracy comparable to state-of-the-art methods, suitable for real-time applications.

Contribution

The paper presents EDANet, a new efficient dense module with asymmetric convolution that significantly improves inference speed without sacrificing accuracy.

Findings

EDANet is 2.7 times faster than ICNet.

EDANet achieves similar mIoU scores to existing methods.

EDANet runs at 108 FPS on high-resolution inputs.

Abstract

Real-time semantic segmentation plays an important role in practical applications such as self-driving and robots. Most semantic segmentation research focuses on improving estimation accuracy with little consideration on efficiency. Several previous studies that emphasize high-speed inference often fail to produce high-accuracy segmentation results. In this paper, we propose a novel convolutional network named Efficient Dense modules with Asymmetric convolution (EDANet), which employs an asymmetric convolution structure and incorporates dilated convolution and dense connectivity to achieve high efficiency at low computational cost and model size. EDANet is 2.7 times faster than the existing fast segmentation network, ICNet, while it achieves a similar mIoU score without any additional context module, post-processing scheme, and pretrained model. We evaluate EDANet on Cityscapes and CamVid datasets, and compare it with the other state-of-art systems. Our network can run with the high-resolution inputs at the speed of 108 FPS on one GTX 1080Ti.