DRE-Net: A Dynamic Radius-Encoding Neural Network with an Incremental Training Strategy for Interactive Segmentation of Remote Sensing Images

Abstract

Semantic segmentation of remote sensing (RS) images, which is a fundamental research topic, classifies each pixel in an image. It plays an essential role in many downstream RS areas, such as land-cover mapping, road extraction, traffic monitoring, and so on. Recently, although deep-learning-based methods have shown their dominance in automatic semantic segmentation of RS imagery, the performance of these existing methods has relied heavily on large amounts of high-quality training data, which are usually hard to obtain in practice. Moreover, human-in-the-loop semantic segmentation of RS imagery cannot be completely replaced by automatic segmentation models, since automatic models are prone to error in some complex scenarios. To address these issues, in this paper, we propose an improved, smart, and interactive segmentation model, DRE-Net, for RS images. The proposed model facilitates humans’ performance of segmentation by simply clicking a mouse. Firstly, a dynamic radius-encoding (DRE) algorithm is designed to distinguish the purpose of each click, such as a click for the selection of a segmentation outline or for fine-tuning. Secondly, we propose an incremental training strategy to cause the proposed model not only to converge quickly, but also to obtain refined segmentation results. Finally, we conducted comprehensive experiments on the Potsdam and Vaihingen datasets and achieved 9.75% and 7.03% improvements in NoC95 compared to the state-of-the-art results, respectively. In addition, our DRE-Net can improve the convergence and generalization of a network with a fast inference speed.

ARE-Net: An Improved Interactive Model for Accurate Building Extraction in High-Resolution Remote Sensing Imagery

“…It performed region segmentation in the click area and refined the segmented area, significantly reducing the computation time and number of model parameters. To further enhance the performance of the segmentation model, dynamic encoding and phased incremental learning strategies were proposed [49]. Interactive building extraction techniques can make full use of user knowledge to guide the extraction process and refine the extraction accuracy.…”

“…In the later stage, the focus of network training should be shifted to finetuning the mask output by the network according to the correction information provided by each interaction, which can also weaken the impact of clicks to prevent convergence (CG) deterioration caused by misclassification or blurring and improve the convergence speed. Some methods, such as DRE-Net [49], employ an incremental learning training strategy to enhance the low rate of convergence (RoC), but this increases the training cost. Furthermore, the interaction features obtained from fixed-radius clicks overlook the fine-tuning effect of correction information and lead to confusion in the building extraction task for high-resolution remote sensing images.…”