HSI-UNITED
research landscape. Despite decades of progress in sensor development and application-specific modelling, HSI remains divided across
domains, instruments, and methodologies. This project establishes a unified framework — combining representation learning, spectral
world models, transfer learning, and federated computation — to enable camera- and domain-agnostic analysis of spectral data.
Project description
The research advances fundamental understanding of how spectral information behaves under physical, environmental, and sensor-induced transformations, creating models that can learn, adapt, and generalise across heterogeneous data sources without sharing raw data. The framework is implemented through three complementary dimensions: United Methods (developing camera-agnostic world models), United Knowledge (integrating cross-domain scientific understanding), and United Scientists (building an open, federated collaboration network).
Through an extensive network of engaged international collaborators of leading HSI researchers, laboratories, and world-leading HS imager manufacturers, HSI-UNITED bridges methodological gaps and fosters open, community-driven research practices. The outcome will be a reproducible and privacy-preserving computational foundation for next-generation imaging science — uniting methods, knowledge, and scientists under a shared spectral vision.
