CascadePSP: Toward Class-Agnostic and Very High-Resolution Segmentation via Global and Local Refinement

State-of-the-art semantic segmentation methods were almost exclusively trained on images within a fixed resolution range. These segmentations are inaccurate for very high-resolution images since using bicubic upsampling of low-resolution segmentation does not adequately capture high-resolution details along object boundaries. In this paper, we propose a novel approach to address the high-resolution segmentation problem without using any high-resolution training data. The key insight is our CascadePSP network which refines and corrects local boundaries whenever possible. Although our network is trained with low-resolution segmentation data, our method is applicable to any resolution even for very high-resolution images larger than 4K. We present quantitative and qualitative studies on different datasets to show that CascadePSP can reveal pixel-accurate segmentation boundaries using our novel refinement module without any finetuning. Thus, our method can be regarded as class-agnostic. Finally, we demonstrate the application of our model to scene parsing in multi-class segmentation.

PDF Abstract CVPR 2020 PDF CVPR 2020 Abstract

Datasets


Introduced in the Paper:

BIG

Used in the Paper:

ADE20K DeepGlobe

Results from the Paper


 Ranked #1 on Semantic Segmentation on BIG (using extra training data)

     Get a GitHub badge
Task Dataset Model Metric Name Metric Value Global Rank Uses Extra
Training Data
Result Benchmark
Semantic Segmentation BIG PSPNet + CascadePSP IoU 93.93 # 1
mBA 75.32 # 1
Semantic Segmentation BIG DeepLabV3+ + CascadePSP IoU 92.23 # 3
mBA 74.59 # 3
Semantic Segmentation BIG RefineNet + CascadePSP IoU 92.79 # 2
mBA 74.77 # 2
Semantic Segmentation BIG FCN + CascadePSP IoU 77.87 # 4
mBA 67.04 # 4
Land Cover Classification DeepGlobe CascadePSP mIOU 68.5 # 5

Methods