Method Details

Details for method 'TuSimple_Coarse'

name	TuSimple_Coarse
challenge	pixel-level semantic labeling
details	Here we show how to improve pixel-wise semantic segmentation by manipulating convolution-related operations that are better for practical use. First, we implement dense upsampling convolution (DUC) to generate pixel-level prediction, which is able to capture and decode more detailed information that is generally missing in bilinear upsampling. Second, we propose a hybrid dilated convolution (HDC) framework in the encoding phase. This framework 1) effectively enlarges the receptive fields of the network to aggregate global information; 2) alleviates what we call the "gridding issue" caused by the standard dilated convolution operation. We evaluate our approaches thoroughly on the Cityscapes dataset, and achieve a new state-of-art result of 80.1% mIOU in the test set. We also are state-of-the-art overall on the KITTI road estimation benchmark and the PASCAL VOC2012 segmentation task. Pretrained models are available at https://goo.gl/DQMeun.
publication	Understanding Convolution for Semantic Segmentation Panqu Wang, Pengfei Chen, Ye Yuan, Ding Liu, Zehua Huang, Xiaodi Hou, Garrison Cottrell https://arxiv.org/abs/1702.08502
project page / code	https://github.com/TuSimple/TuSimple-DUC
used Cityscapes data	fine annotations, coarse annotations
used external data	ImageNet
runtime	n/a
subsampling	no
submission date	February, 2017
previous submissions	1, 2, 3, 4