Published on Sat Jul 13 2019

Adaptive Context Encoding Module for Semantic Segmentation

Congcong Wang, Faouzi Alaya Cheikh, Azeddine Beghdadi, Ole Jakob Elle

The object sizes in images are diverse, therefore, capturing multiple scale information is essential for semantic segmentation. Existing context aggregation methods such as pyramid pooling module (PPM) and atrous spatialPyramid pooling (ASPP) design different pooling size or atrous rate.

0
0
0
Abstract

The object sizes in images are diverse, therefore, capturing multiple scale context information is essential for semantic segmentation. Existing context aggregation methods such as pyramid pooling module (PPM) and atrous spatial pyramid pooling (ASPP) design different pooling size or atrous rate, such that multiple scale information is captured. However, the pooling sizes and atrous rates are chosen manually and empirically. In order to capture object context information adaptively, in this paper, we propose an adaptive context encoding (ACE) module based on deformable convolution operation to argument multiple scale information. Our ACE module can be embedded into other Convolutional Neural Networks (CNN) easily for context aggregation. The effectiveness of the proposed module is demonstrated on Pascal-Context and ADE20K datasets. Although our proposed ACE only consists of three deformable convolution blocks, it outperforms PPM and ASPP in terms of mean Intersection of Union (mIoU) on both datasets. All the experiment study confirms that our proposed module is effective as compared to the state-of-the-art methods.

Sat Mar 21 2020
Computer Vision
BiCANet: Bi-directional Contextual Aggregating Network for Image Semantic Segmentation
Bi-directional Contextual Aggregating Network (BiCANet) for semantic segmentation. Unlike previous approaches that encode contexts in feature space, BiCANet aggregates contextual cues from a categorical perspective.
0
0
0
Thu Feb 27 2020
Computer Vision
Attention-guided Chained Context Aggregation for Semantic Segmentation
The way features propagate in Fully Convolutional Networks is of momentous importance to capture multi-scale contexts. This paper proposes a novel series-parallel hybrid paradigm to diversify feature propagation.
0
0
0
Fri Mar 23 2018
Computer Vision
Context Encoding for Semantic Segmentation
The proposed Context Encoding Module significantly improves semantic segmentation results. Our approach has achieved new state-of-the-art results 51.7% mIoU on PASCAL-Context. Our 14 layer network has achieved an error rate of 3.45%.
0
0
0
Sat Mar 20 2021
Computer Vision
A Novel Upsampling and Context Convolution for Image Semantic Segmentation
0
0
0
Tue Jun 08 2021
Computer Vision
SpaceMeshLab: Spatial Context Memoization and Meshgrid Atrous Convolution Consensus for Semantic Segmentation
Spatial Context Memoization (SpaM) is a bypassing branch for spatial context. SpaM retains the input dimension andcommunicates its spatial context and rich semantic information mutually with the backbone network. SpaceMeshLab achieves 82.0% mIoU in Cityscapes
0
0
0
Fri Apr 17 2020
Computer Vision
Learning to Predict Context-adaptive Convolution for Semantic Segmentation
Long-range contextual information is essential for achieving high-performance semantic segmentation. In CaC-Net, a set of context-adaptive convolution kernels are predicted from the global contextual information in a parameter-efficient manner.
0
0
0