A Change Detection Reality Check
This is an incremental study that questions the claimed advancements in change detection methods for remote sensing researchers.
The paper tackled the problem of evaluating progress in change detection deep learning architectures in remote sensing, finding that a simple U-Net segmentation baseline without advanced tricks remains a top performer.
In recent years, there has been an explosion of proposed change detection deep learning architectures in the remote sensing literature. These approaches claim to offer state-of-the-art performance on different standard benchmark datasets. However, has the field truly made significant progress? In this paper we perform experiments which conclude a simple U-Net segmentation baseline without training tricks or complicated architectural changes is still a top performer for the task of change detection.