Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Generalized Few-Shot Semantic Segmentation in Remote Sensing: Challenge and Benchmark

Clifford Broni-Bediako, Junshi Xia, Jian Song, Hongruixuan Chen, Mennatullah Siam, Naoto Yokoya

TL;DR

This work is the first to propose a generalized few-shot segmentation benchmark for remote sensing and releases the dataset augmenting OpenEarthMap (OEM) with additional classes labeled for the generalized few-shot evaluation setting.

Abstract

Learning with limited labelled data is a challenging problem in various applications, including remote sensing. Few-shot semantic segmentation is one approach that can encourage deep learning models to learn from few labelled examples for novel classes not seen during the training. The generalized few-shot segmentation setting has an additional challenge which encourages models not only to adapt to the novel classes but also to maintain strong performance on the training base classes. While previous datasets and benchmarks discussed the few-shot segmentation setting in remote sensing, we are the first to propose a generalized few-shot segmentation benchmark for remote sensing. The generalized setting is more realistic and challenging, which necessitates exploring it within the remote sensing context. We release the dataset augmenting OpenEarthMap with additional classes labelled for the generalized few-shot evaluation setting. The dataset is released during the OpenEarthMap land cover mapping generalized few-shot challenge in the L3D-IVU workshop in conjunction with CVPR 2024. In this work, we summarize the dataset and challenge details in addition to providing the benchmark results on the two phases of the challenge for the validation and test sets.

Generalized Few-Shot Semantic Segmentation in Remote Sensing: Challenge and Benchmark

TL;DR

This work is the first to propose a generalized few-shot segmentation benchmark for remote sensing and releases the dataset augmenting OpenEarthMap (OEM) with additional classes labeled for the generalized few-shot evaluation setting.

Abstract

Learning with limited labelled data is a challenging problem in various applications, including remote sensing. Few-shot semantic segmentation is one approach that can encourage deep learning models to learn from few labelled examples for novel classes not seen during the training. The generalized few-shot segmentation setting has an additional challenge which encourages models not only to adapt to the novel classes but also to maintain strong performance on the training base classes. While previous datasets and benchmarks discussed the few-shot segmentation setting in remote sensing, we are the first to propose a generalized few-shot segmentation benchmark for remote sensing. The generalized setting is more realistic and challenging, which necessitates exploring it within the remote sensing context. We release the dataset augmenting OpenEarthMap with additional classes labelled for the generalized few-shot evaluation setting. The dataset is released during the OpenEarthMap land cover mapping generalized few-shot challenge in the L3D-IVU workshop in conjunction with CVPR 2024. In this work, we summarize the dataset and challenge details in addition to providing the benchmark results on the two phases of the challenge for the validation and test sets.
Paper Structure (14 sections, 3 figures, 2 tables)

This paper contains 14 sections, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Satellite Imagery Datasets
  4. Few-Shot Learning
  5. Dataset
  6. Data Curation and Annotation
  7. Classes and Data Split
  8. New classes
  9. OpenEarthMap classes
  10. Dataset splits
  11. Challenge and Benchmark
  12. Challenge Details
  13. Challenge Results
  14. Conclusion

Figures (3)

  • Figure 1: The number of images and geographical regions in the OEM-GFSS dataset across the six continents. There are 408 images from 73 geographical regions. OEM-GFSS has a greater representation in Europe and less in Oceania and North America.
  • Figure 2: OEM-GFSS validation set examples. The first two columns: examples of novel classes in the support set. The second two columns: base and novel classes in the query set.
  • Figure 3: Examples of visual land cover mapping results of the baseline model on the test set of the OEM-GFSS dataset. (a) is novel classes of the test set and (b) is base classes. Query images can contain both the novel classes and the base classes, and all the classes are to be recognised.