MFDS-Net: Multi-Scale Feature Depth-Supervised Network for Remote Sensing Change Detection with Global Semantic and Detail Information

TL;DR

MFDS-Net tackles remote sensing change detection by jointly preserving edge details and capturing global semantic information. It introduces three novel modules—MDPM for detail preservation, GSEM for global context, and DFIM for differential feature integration—built on a DO-Conv enhanced backbone with deep supervision. The method delivers state-of-the-art results on LEVIR-CD, WHU-CD, and GZ-CD, with robust performance under challenging lighting and color variation. These improvements translate to more accurate building-change delineation and stronger generalization, with potential applicability to cross-domain segmentation tasks.

Abstract

Change detection as an interdisciplinary discipline in the field of computer vision and remote sensing at present has been receiving extensive attention and research. Due to the rapid development of society, the geographic information captured by remote sensing satellites is changing faster and more complex, which undoubtedly poses a higher challenge and highlights the value of change detection tasks. We propose MFDS-Net: Multi-Scale Feature Depth-Supervised Network for Remote Sensing Change Detection with Global Semantic and Detail Information (MFDS-Net) with the aim of achieving a more refined description of changing buildings as well as geographic information, enhancing the localisation of changing targets and the acquisition of weak features. To achieve the research objectives, we use a modified ResNet_34 as backbone network to perform feature extraction and DO-Conv as an alternative to traditional convolution to better focus on the association between feature information and to obtain better training results. We propose the Global Semantic Enhancement Module (GSEM) to enhance the processing of high-level semantic information from a global perspective. The Differential Feature Integration Module (DFIM) is proposed to strengthen the fusion of different depth feature information, achieving learning and extraction of differential features. The entire network is trained and optimized using a deep supervision mechanism. The experimental outcomes of MFDS-Net surpass those of current mainstream change detection networks. On the LEVIR dataset, it achieved an F1 score of 91.589 and IoU of 84.483, on the WHU dataset, the scores were F1: 92.384 and IoU: 86.807, and on the GZ-CD dataset, the scores were F1: 86.377 and IoU: 76.021. The code is available at https://github.com/AOZAKIiii/MFDS-Net

Figures (24)

• Figure 1: The results of the experiments performed on three datasets, the first sample set is the LEVIR-CD dataset, the second sample set is the WHU-CD dataset and the third is the GZ-CD dataset. The last column shows the heat map obtained from Ours.
• Figure 2: MFDS-Net master network diagram, $\text{Re}sNet_{34}$ as the backbone network completes feature extraction. Feature enhancement of feature information at different scales is performed by MDPM. GSEM accomplishes the integration of contextual feature information from a global viewpoint. DFIM highlights difference features. Enhancing the network's training procedure through the implementation of a deep supervision mechanism.
• Figure 3: (a) shows the specific process of feature reconstruction. (b) shows the traditional convolution process. (c) shows the Depthwise convolution process.
• Figure 4: The DO-Conv process.
• Figure 5: Multi-scale Detail Preservation Module(MDPM).