Weakly-Supervised Domain Adaptation With Adversarial Entropy for Building Segmentation in Cross-Domain Aerial Imagery

Abstract

Building segmentation is a classical and challenging task in high-resolution remote sensing imagery. This approach has achieved remarkable performance based on a fully convolutional network (FCN) with adequate pixel-wise annotations. However, due to differences in sensor technology as well as appearance in different regions, datasets gathered from these various sources are quite distinct, and dense annotations for a particular area are not always available. Thus, directly applying a segmentation model trained on one dataset (source domain) to another unseen dataset (target domain) usually results in a drop in performance, called the domain gap. In this paper, we propose a weakly-supervised domain adaptation method using adversarial entropy for building segmentation to address this problem. First, we use an adversarial entropy strategy to decrease the entropy and improve the prediction certainty for target images, causing the distributions between the source and target domains to become closer to each other. Second, we propose a simple and effective self-training strategy for the target domain that produces high-confidence predictions using pseudo-labels. We use a series of thresholds to generate the pseudo-labels without introducing extra parameters. This strategy effectively enhances the discriminability of the target domain and further minimizes the distribution discrepancy between the two domains. Experiments on cross-domain aerial datasets have demonstrated the effectiveness and superiority of our proposed method when compared to other state-of-the-art methods.

Semi-Supervised Building Footprint Generation With Feature and Output Consistency Training

“…Afterward, the generator outputs the segmentation masks of unlabeled images, while the discriminator distinguishes trustworthy regions in their predicted results to provide additional supervisory signals. Considering that the adversarial training strategy may be insufficient to guide network training, pseudo labels are generated by selecting high-confident segmentation predictions for unlabeled images [48]. Afterward, pseudo-building masks are incorporated to expand the training data and the generator is retrained.…”

Enhanced Semantic Information Transfer of Multi-Domain Samples: An Adversarial Edge Detection Method Using Few High-Resolution Remote Sensing Images

“…Song et al [ 35 ] designed a subspace alignment module to add to the CCN model, which alleviated the domain distribution discrepancy and somewhat solved the problem of different domain samples in scene classification. Yao et al [ 36 ] proposed a weakly supervised domain adaptation method by utilizing adversarial entropy, which addresses the domain gap problem in building semantic segmentation by using an adversarial entropy strategy and a self-training strategy. However, these methods for nonedge-dominated tasks, such as classification tasks and semantic segmentation tasks, have less reference to edge extraction.…”

Unsupervised Building Extraction from Multimodal Aerial Data Based on Accurate Vegetation Removal and Image Feature Consistency Constraint

“…However, deep-learning-based methods still rely on a large number of labelled samples to obtain satisfactory results, and these samples are often manually labelled, which is time and labour consuming. Furthermore, if the test dataset and the training dataset differ greatly, the deep learning models may perform poorly [25][26][27], and new samples are often manually labelled to overcome this problem. Ideally, we would establish a very large dataset to cover as many types of buildings as possible, like the famous ImageNet dataset [28] for image classification tasks.…”