Semantic decoupled representation learning for remote sensing image change detection
This work addresses data insufficiency for remote sensing change detection, but it is incremental as it builds on existing self-supervised learning approaches with a novel disentanglement technique.
The paper tackles the problem of data insufficiency in remote sensing image change detection by proposing a semantic decoupled representation learning method that disentangles representations of different semantic regions using semantic masks. The result shows that the model outperforms ImageNet pre-training, in-domain supervised pre-training, and several recent self-supervised learning methods on two change detection datasets.
Contemporary transfer learning-based methods to alleviate the data insufficiency in change detection (CD) are mainly based on ImageNet pre-training. Self-supervised learning (SSL) has recently been introduced to remote sensing (RS) for learning in-domain representations. Here, we propose a semantic decoupled representation learning for RS image CD. Typically, the object of interest (e.g., building) is relatively small compared to the vast background. Different from existing methods expressing an image into one representation vector that may be dominated by irrelevant land-covers, we disentangle representations of different semantic regions by leveraging the semantic mask. We additionally force the model to distinguish different semantic representations, which benefits the recognition of objects of interest in the downstream CD task. We construct a dataset of bitemporal images with semantic masks in an effortless manner for pre-training. Experiments on two CD datasets show our model outperforms ImageNet pre-training, in-domain supervised pre-training, and several recent SSL methods.