![[2509.13229] Curriculum Multi-Task Self-Supervision Improves Lightweight Architectures for Onboard Satellite Hyperspectral Image Segmentation](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2509.13229] Curriculum Multi-Task Self-Supervision Improves Lightweight Architectures for Onboard Satellite Hyperspectral Image Segmentation
Summary
This article presents a novel framework, Curriculum Multi-Task Self-Supervision Learning (CMTSSL), aimed at enhancing lightweight architectures for hyperspectral image segmentation in satellite applications.
Why It Matters
With the increasing demand for efficient onboard processing of hyperspectral images, this research offers a significant advancement in machine learning techniques that can improve data handling and analysis in satellite systems, which is crucial for environmental monitoring and land-use classification.
Key Takeaways
- CMTSSL integrates masked image modeling with spatial and spectral jigsaw puzzle solving.
- The framework improves lightweight model performance for hyperspectral image segmentation.
- Results show consistent gains across four public benchmark datasets.
- CMTSSL is over 16,000 times lighter than some state-of-the-art models.
- The approach enhances representation learning for real-world applications.
Computer Science > Computer Vision and Pattern Recognition arXiv:2509.13229 (cs) [Submitted on 16 Sep 2025 (v1), last revised 16 Feb 2026 (this version, v2)] Title:Curriculum Multi-Task Self-Supervision Improves Lightweight Architectures for Onboard Satellite Hyperspectral Image Segmentation Authors:Hugo Carlesso, Josiane Mothe, Radu Tudor Ionescu View a PDF of the paper titled Curriculum Multi-Task Self-Supervision Improves Lightweight Architectures for Onboard Satellite Hyperspectral Image Segmentation, by Hugo Carlesso and 2 other authors View PDF HTML (experimental) Abstract:Hyperspectral imaging (HSI) captures detailed spectral signatures across hundreds of contiguous bands per pixel, being indispensable for remote sensing applications such as land-cover classification, change detection, and environmental monitoring. Due to the high dimensionality of HSI data and the slow rate of data transfer in satellite-based systems, compact and efficient models are required to support onboard processing and minimize the transmission of redundant or low-value data. To this end, we introduce a novel curriculum multi-task self-supervised learning (CMTSSL) framework designed for lightweight architectures for HSI analysis. CMTSSL integrates masked image modeling with decoupled spatial and spectral jigsaw puzzle solving, guided by a curriculum learning strategy that progressively increases data difficulty during self-supervision. This enables the encoder to jointly capture fine-grained...