Abstract:Semantic segmentation of remote sensing images (RSIs) is effective for large-scale land cover mapping, which heavily relies on a large amount of training data with laborious pixel-level labeling. Due to the easy availability of image-level labels, weakly supervised semantic segmentation (WSSS) based on them has attracted intensive attention. However, existing image-level WSSS methods for RSIs mainly focus on binary segmentation, which are difficult to apply to multiclass scenarios. This study proposes a comprehensive framework for image-level multiclass WSSS of RSIs, consisting of appropriate image-level label generation, high-quality pixel-level pseudo mask generation, and segmentation network iterative training. Specifically, a training sample filtering method, as well as a dataset co-occurrence evaluation metric, is proposed to demonstrate proper image-level training samples. Leveraging multiclass class activation maps (CAMs), an uncertainty-driven pixel-level weighted mask is proposed to relieve the overfitting of labeling noise in pseudo masks when training the segmentation network. Extensive experiments demonstrate that the proposed framework can achieve high-quality multiclass WSSS performance with image-level labels, which can attain 94.23% and 90.77% of the mean intersection over union (mIoU) from pixel-level labels for the ISPRS Potsdam and Vaihingen datasets, respectively. Beyond that, for the DeepGlobe dataset with more complex landscapes, the WSSS framework can achieve an accuracy close to 99% of the fully supervised case. In addition, we further demonstrate that compared to adopting multiple binary WSSS models, directly training a multiclass WSSS model can achieve better results, which can provide new thoughts to achieve WSSS of RSIs for multiclass application scenarios. Our code is publically available at https://github.com/NJU-LHRS/OME.

Sparse and Stratified Sampling Method for One - Class Classification of Remote Sensing Images

On the Sampling Strategies for Evaluation of Joint Spectral-Spatial Information Based Classifiers

Active One-Class Classification of Remote Sensing Image

On the Sampling Strategy for Evaluation of Spectral-Spatial Methods in Hyperspectral Image Classification

One-Class Classification of Remote Sensing Images Using Kernel Sparse Representation.

A New Robust Approach for Remote Sensing Image Regional Classification

Dense v.s. Sparse: A Comparative Study of Sampling Analysis in Scene Classification of High-Resolution Remote Sensing Imagery

Land-Cover Change Detection Using One-Class Support Vector Machine

Clustering-Based Extraction of Near Border Data Samples for Remote Sensing Image Classification.

One-class Remote Sensing Classification: One-Class Vs. Binary Classifiers

A one-class classification by spatial-contextual for remotely sensed image

Cluster-based Training Data Preselection and Classification for Remote Sensing Images

One-Class Oriented Feature Selection and Classification of Heterogeneous Remote Sensing Images

Multiobjective-Based Sparse Representation Classifier for Hyperspectral Imagery Using Limited Samples

SVDD-based one-class land-cover mapping using optimal training samples

A Self-Adaptive Selective Method of Remote Sensing Image Classification Algorithms

High Resolution Remotely Sensed Imagery Classification For Urban Environment Monitoring Based On Support Vector Machine

One model is enough: Toward multiclass weakly supervised remote sensing image semantic segmentation

Remote Sensing Image Classification: A survey of support-vector-machine-based advanced techniques

A Generalized Image Scene Decomposition-Based System For Supervised Classification Of Very High Resolution Remote Sensing Imagery

A remote sensing image classification method based on sparse representation