🤖 AI Summary
To address the limited pseudo-label diversity, model redundancy, and high GPU memory consumption in semi-supervised learning (SSL) for remote sensing image semantic segmentation, this paper proposes a lightweight and efficient dual-path framework. First, we introduce DiverseHead—a novel multi-decision-head architecture that replaces multiple networks with diverse heads within a single model to generate structurally diverse pseudo-labels. Second, we design DiverseModel—a parallel heterogeneous network architecture that collaboratively produces robust pseudo-labels. Our method integrates consistency regularization, lightweight feature distillation, and adversarial perturbation to significantly improve pseudo-label quality and training stability. Evaluated on four mainstream remote sensing datasets, our approach surpasses state-of-the-art SSL methods. Moreover, DiverseHead serves as a plug-and-play module compatible with various SSL frameworks, reducing GPU memory usage by over 40% and accelerating inference by 35%.
📝 Abstract
Semi-supervised learning (SSL) aims to help reduce the cost of the manual labelling process by leveraging a substantial pool of unlabelled data alongside a limited set of labelled data during the training phase. Since pixel-level manual labelling in large-scale remote sensing imagery is expensive and time-consuming, semi-supervised learning has become a widely used solution to deal with this. However, the majority of existing SSL frameworks, especially various teacher-student frameworks, are too bulky to run efficiently on a GPU with limited memory. There is still a lack of lightweight SSL frameworks and efficient perturbation methods to promote the diversity of training samples and enhance the precision of pseudo labels during training. In order to fill this gap, we proposed a simple, lightweight, and efficient SSL architecture named extit{DiverseHead}, which promotes the utilisation of multiple decision heads instead of multiple whole networks. Another limitation of most existing SSL frameworks is the insufficient diversity of pseudo labels, as they rely on the same network architecture and fail to explore different structures for generating pseudo labels. To solve this issue, we propose extit{DiverseModel} to explore and analyse different networks in parallel for SSL to increase the diversity of pseudo labels. The two proposed methods, namely extit{DiverseHead} and extit{DiverseModel}, both achieve competitive semantic segmentation performance in four widely used remote sensing imagery datasets compared to state-of-the-art semi-supervised learning methods. Meanwhile, the proposed lightweight DiverseHead architecture can be easily applied to various state-of-the-art SSL methods while further improving their performance. The code is available at https://github.com/WANLIMA-CARDIFF/DiverseNet.
