Zero-Shot Multi-Spectral Learning: Reimagining a Generalist Multimodal Gemini 2.5 Model for Remote Sensing Applications

Ganesh Mallya Yotam Gigi Dahun Kim Maxim Neumann, Genady Beryozkin Tomer Shekel Anelia Angelova · 2025

Read Zero-Shot Multi-Spectral Learning: Reimagining a Generalist Multimodal Gemini 2.5 Model for Remote Sensing Applications by Ganesh Mallya Yotam Gigi Dahun Kim Maxim Neumann, Genady Beryozkin Tomer Shekel Anelia Angelova (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2509-19087.txt.

Description

Satellite images that see beyond visible light can identify ice, water, and crops far better than ordinary photos. Ganesh Mallya and six co-authors ask whether a general-purpose AI model, trained only on normal images, can read such multi-spectral data with no extra training. Specialist models for this data are costly to build and maintain. Big general models, meanwhile, cannot take the extra bands as input at all.

Their training-free approach reshapes the spectral data into a form the general model can use, and adds written instructions carrying expert knowledge. They test it with the Gemini 2.5 model on standard land cover and land use benchmarks. The method scored strong zero-shot gains there. The authors argue that specialists in remote sensing can now tap large general models for new sensor data with little setup.

Catalog record

CREATORGanesh Mallya Yotam Gigi Dahun Kim Maxim Neumann, Genady Beryozkin Tomer Shekel Anelia Angelova
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE37 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2509-19087

Browse the full library · About the Tower of Babel