Spectral-Spatial Contrastive Learning Framework for Regression on Hyperspectral Data
This work addresses a gap in hyperspectral data analysis for regression tasks, though it appears incremental as it builds on existing contrastive learning and backbone architectures.
The authors tackled the shortage of contrastive learning methods for regression tasks on hyperspectral data by proposing a spectral-spatial framework and relevant transformations, which significantly improved the performance of backbone models like 3D CNNs and transformers on synthetic and real datasets.
Contrastive learning has demonstrated great success in representation learning, especially for image classification tasks. However, there is still a shortage in studies targeting regression tasks, and more specifically applications on hyperspectral data. In this paper, we propose a spectral-spatial contrastive learning framework for regression tasks for hyperspectral data, in a model-agnostic design allowing to enhance backbones such as 3D convolutional and transformer-based networks. Moreover, we provide a collection of transformations relevant for augmenting hyperspectral data. Experiments on synthetic and real datasets show that the proposed framework and transformations significantly improve the performance of all studied backbone models.