Detecting Building Changes with Off-Nadir Aerial Images

TL;DR

This paper introduces MTGCD-Net, a multi-task learning model that improves building change detection in off-nadir aerial images by addressing challenges like misalignment and semantic ambiguity, and it is validated on a new dataset BANDON.

Contribution

The paper proposes a novel multi-task guided change detection network and a new benchmark dataset for off-nadir aerial images, enhancing accuracy in building change detection.

Findings

MTGCD-Net outperforms previous methods on the BANDON dataset.

Auxiliary tasks improve building parsing and matching accuracy.

The model effectively handles roof misalignment and semantic ambiguity.

Abstract

The tilted viewing nature of the off-nadir aerial images brings severe challenges to the building change detection (BCD) problem: the mismatch of the nearby buildings and the semantic ambiguity of the building facades. To tackle these challenges, we present a multi-task guided change detection network model, named as MTGCD-Net. The proposed model approaches the specific BCD problem by designing three auxiliary tasks, including: (1) a pixel-wise classification task to predict the roofs and facades of buildings; (2) an auxiliary task for learning the roof-to-footprint offsets of each building to account for the misalignment between building roof instances; and (3) an auxiliary task for learning the identical roof matching flow between bi-temporal aerial images to tackle the building roof mismatch problem. These auxiliary tasks provide indispensable and complementary building parsing and matching information. The predictions of the auxiliary tasks are finally fused to the main building change detection branch with a multi-modal distillation module. To train and test models for the BCD problem with off-nadir aerial images, we create a new benchmark dataset, named BANDON. Extensive experiments demonstrate that our model achieves superior performance over the previous state-of-the-art competitors.