Multi-scale feature progressive fusion network for remote sensing image change detection

Abstract

Presently, research on deep learning-based change detection (CD) methods has become a hot topic. In particular, feature pyramid networks (FPNs) are widely used in CD tasks to gradually fuse semantic features. However, existing FPN-based CD methods do not correctly detect the complete change region and cannot accurately locate the boundaries of the change region. To solve these problems, a new Multi-Scale Feature Progressive Fusion Network (MFPF-Net) is proposed, which consists of three innovative modules: Layer Feature Fusion Module (LFFM), Multi-Scale Feature Aggregation Module (MSFA), and Multi-Scale Feature Distribution Module (MSFD). Specifically, we first concatenate the features of each layer extracted from the bi-temporal images with their difference maps, and the resulting change maps fuse richer semantic information while effectively representing change regions. Then, the obtained change maps of each layer are directly aggregated, which improves the effective communication and full fusion of feature maps in CD while avoiding the interference of indirect information. Finally, the aggregated feature maps are layered again by pooling and convolution operations, and then a feature fusion strategy with a pyramid structure is used, with layers fused from low to high, to obtain richer contextual information, so that each layer of the layered feature maps has original semantic information and semantic features of other layers. We conducted comprehensive experiments on three publicly available benchmark datasets, CDD, LEVIR-CD, and WHU-CD to verify the effectiveness of the method, and the experimental results show that the method in this paper outperforms other comparative methods.

VisionTwinNet: Gated Clarity Enhancement Paired With Light-Robust CD Transformers

“…And the comparison between (2) and ( 5) in FC-Siam-Di [22] FC-Siam-Conc [22] DTCDSCN [9] STANet [2] IFNet [10] SNUNet [14] BIT…”

“…At present, various CD algorithms have achieved high precision on mainstream datasets such as DSIFN, LEVIR, and WHU-CD [13], [14], [15], [16]. However, these algorithms exhibit limitations when dealing with bitemporal images under varying illumination conditions and noise.…”

YOLO-EMS: A Lightweight Object Detection Algorithm for Remote Sensing Images

“…Object detection is a trivial task for humans, but over a decade ago, having computers do it was a very challenging task. However, with the development of deep learning, computer vision technology has been widely used in multiple fields such as intelligent security 1 , autonomous driving, remote sensing detection 2,3 , medical and pharmaceutical 4 , agriculture 5 , intelligent transportation 6 , and information security 7 . The core task of object detection is to recognize and locate all instances of objects in the field of view (such as humans, dogs, cars, and tables).…”

Cross‐attention neural network for land cover change detection with remote sensing images

“…In addition, a supervised attention network was proposed for achieving a satisfactory detection map, such as deeply supervised attention metric‐based networks (DSAMNet) (Shi et al., 2022) and deeply supervised attentive high‐resolution networks (Wang et al., 2021). Siamese‐based spatial–temporal attention neural networks (Chen & Shi, 2020), cross‐temporal interaction symmetric attention networks (Lu et al., 2022), and hierarchy metric learning networks with dual attention have been widely used for LCCD with RSIs as well.…”