The majority of remote sensing image change detection models focus on extracting high-level semantic features. However, it is difficult to detect the changing area with large differences in size simultaneously, and the accurate changing area edge is difficult. To solve these problems, we propose a weighted rich-scale inception coder network (WRICNet), consisting of the weighted rich-scale inception module and the weighted rich-scale coder module. The former can retain the low-level multiscale feature (LMF), and the latter can extract the high-level multiscale feature (HMF). By fusing LMF and HMF, both large and small changing areas can be detected and make changing area edge accurate. To complement LMF and HMF well, we propose a weighted scale block, which assigns appropriate weights to multiscale features. Compared to comparative methods, performance experiments on datasets demonstrate that our proposed method can further reduce false alarms and miss alarms and make the changing area edge more accurate. Furthermore, ablation studies show that our training strategy, model settings, and improvements of the inception and rich-scale blocks are effective.
Existing 3D Gaussian (3DGS) based methods tend to produce blurriness and artifacts on delicate textures (small objects and high-frequency textures) in aerial large-scale scenes. The reason is that the delicate textures usually occupy a relatively small number of pixels, and the accumulated gradients from loss function are difficult to promote the splitting of 3DGS. To minimize the rendering error, the model will use a small number of large Gaussians to cover these details, resulting in blurriness and artifacts. To solve the above problem, we propose a novel hierarchical Gaussian: JumpingGS. JumpingGS assigns different levels to Gaussians to establish a hierarchical representation. Low-level Gaussians are responsible for the coarse appearance, while high-level Gaussians are responsible for the details. First, we design a splitting strategy that allows low-level Gaussians to skip intermediate levels and directly split the appropriate high-level Gaussians for delicate textures. This level-jump splitting ensures that the weak gradients of delicate textures can always activate a higher level instead of being ignored by the intermediate levels. Second, JumpingGS reduces the gradient and opacity thresholds for density control according to the representation levels, which improves the sensitivity of high-level Gaussians to delicate textures. Third, we design a novel training strategy to detect training views in hard-to-observe regions, and train the model multiple times on these views to alleviate underfitting. Experiments on aerial large-scale scenes demonstrate that JumpingGS outperforms existing 3DGS-based methods, accurately and efficiently recovering delicate textures in large scenes.