Method Details

Details for method 'SPGNet'


Method overview

name SPGNet
challenge pixel-level semantic labeling
details Multi-scale context module and single-stage encoder-decoder structure are commonly employed for semantic segmentation. The multi-scale context module refers to the operations to aggregate feature responses from a large spatial extent, while the single-stage encoder-decoder structure encodes the high-level semantic information in the encoder path and recovers the boundary information in the decoder path. In contrast, multi-stage encoder-decoder networks have been widely used in human pose estimation and show superior performance than their single-stage counterpart. However, few efforts have been attempted to bring this effective design to semantic segmentation. In this work, we propose a Semantic Prediction Guidance (SPG) module which learns to re-weight the local features through the guidance from pixel-wise semantic prediction. We find that by carefully re-weighting features across stages, a two-stage encoder-decoder network coupled with our proposed SPG module can significantly outperform its one-stage counterpart with similar parameters and computations. Finally, we report experimental results on the semantic segmentation benchmark Cityscapes, in which our SPGNet attains 81.1% on the test set using only 'fine' annotations.
publication SPGNet: Semantic Prediction Guidance for Scene Parsing
Bowen Cheng, Liang-Chieh Chen, Yunchao Wei, Yukun Zhu, Zilong Huang, Jinjun Xiong, Thomas Huang, Wen-Mei Hwu, Honghui Shi
ICCV 2019
project page / code
used Cityscapes data fine annotations
used external data ImageNet
runtime n/a
subsampling no
submission date March, 2019
previous submissions


Average results

Metric Value
IoU Classes 81.0901
iIoU Classes 61.447
IoU Categories 92.1467
iIoU Categories 82.0908


Class results

Class IoU iIoU
road 98.7924 -
sidewalk 87.605 -
building 93.7729 -
wall 56.4799 -
fence 61.9184 -
pole 71.8958 -
traffic light 79.9507 -
traffic sign 82.0779 -
vegetation 94.0825 -
terrain 73.5125 -
sky 96.099 -
person 88.6817 73.4989
rider 74.9341 54.8083
car 96.4726 91.5607
truck 67.3384 42.5601
bus 84.8188 57.6436
train 81.7982 53.6576
motorcycle 71.1094 51.2846
bicycle 79.3718 66.5625


Category results

Category IoU iIoU
flat 98.7939 -
nature 93.7539 -
object 77.4615 -
sky 96.099 -
construction 94.2022 -
human 88.8411 74.4925
vehicle 95.8751 89.6891



Download results as .csv file

Benchmark page