Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

FastFCN: Rethinking Dilated Convolution in the Backbone for Semantic Segmentation

About

Modern approaches for semantic segmentation usually employ dilated convolutions in the backbone to extract high-resolution feature maps, which brings heavy computation complexity and memory footprint. To replace the time and memory consuming dilated convolutions, we propose a novel joint upsampling module named Joint Pyramid Upsampling (JPU) by formulating the task of extracting high-resolution feature maps into a joint upsampling problem. With the proposed JPU, our method reduces the computation complexity by more than three times without performance loss. Experiments show that JPU is superior to other upsampling modules, which can be plugged into many existing approaches to reduce computation complexity and improve performance. By replacing dilated convolutions with the proposed JPU module, our method achieves the state-of-the-art performance in Pascal Context dataset (mIoU of 53.13%) and ADE20K dataset (final score of 0.5584) while running 3 times faster.

Huikai Wu, Junge Zhang, Kaiqi Huang, Kongming Liang, Yizhou Yu• 2019

Related benchmarks

TaskDatasetResultRank
Semantic segmentationADE20K (val)
mIoU44.34
2731
Semantic segmentationPASCAL Context (val)
mIoU53.1
323
Semantic segmentationPascal Context (test)
mIoU51.2
176
Semantic segmentationPASCAL-Context 60 classes (test)
mIoU53.1
54
Semantic segmentationADE20K (test)--
50
Semantic segmentationRELLIS-3D
mIoU0.705
16
Semantic segmentationPascal Context 60 classes w/ background (val)
mIoU53.1
12
Semantic ParsingNYUd2 (test)
mIoU45.4
10
Semantic segmentationADE20K COCO-Place challenge 2017 (test)
Final Score0.5584
7
Image SegmentationADE20K (val)
Pixel Accuracy80.39
5
Showing 10 of 10 rows

Other info

Code

Follow for update