📆 Project Period | |
👤 CIN Visiting Researcher |
Project Summary
Main projects are on Quantum computing for earth observation:
This summary presents an overview of my research within the QC4EO framework, focusing on Quantum Machine Learning for Earth Observation (QML4EO). The work is structured around two main pillars: Applications and Foundations. On the Applications side, QML methods have been developed and tested for a range of EO tasks, including noise filtering in satellite imagery, hybrid quantum-classical models for land classification, object detection, image synthesis with quantum generative models, spatiotemporal forecasting using quantum recurrent networks, and pixel-wise segmentation of geospatial data. These projects demonstrate the potential of quantum-enhanced algorithms to tackle high-dimensional and structured EO data challenges.
The Foundations pillar supports these applications by advancing core quantum learning techniques. Key contributions include the development of quantum neural network architectures such as quanvolutional layers, quantum MLPs, and quantum RNNs, tailored to extract and process features from EO data. Further work has been dedicated to optimizing these models through improved training methods, quantum-informed hyperparameter tuning, and the use of quantum approaches to accelerate neural architecture search. Collectively, these research efforts contribute to building a cohesive and practical framework that integrates quantum technologies into Earth Observation, positioning QML not only as a theoretical innovation but as a viable tool for advancing geospatial intelligence.
For each topic there are one or more publications linked.
Additional Projects are Deep Learning for Earth Observation
This summary outlines my research within the AI4EO framework, emphasizing the use of Deep Learning (DL) to address key environmental and societal challenges through Earth Observation (EO) data. The work is structured into three main thematic areas: climate-related, health-related, and emerging applications.
In the climate-related domain, DL models have been developed for monitoring water resources, predicting solar energy potential from EO-based irradiance data, and estimating air pollution levels through satellite-DNN integration. Additional efforts include early prediction of El Niño events using temporal EO data with recurrent networks, and precipitation forecasting through the fusion of EO and DL for improved weather risk assessments. Health-related applications focus on leveraging EO-derived environmental variables to inform
public health strategies. This includes analyzing links between air quality, temperature, and Covid-19 trends, and modeling dengue outbreak risks using climate indicators and DL-based epidemiological tools.
Other innovative projects include volcanic eruption monitoring with DL applied to remote sensing data, and the development of on-board AI for real-time processing of EO data on satellites. Overall, this body of work demonstrates how DL can transform EO data into actionable insights, supporting timely decision-making across diverse sectors from climate resilience to public health and space-based intelligence.
Development Tools
- Mainly Python as programming language
- Python libraries for AI: Tensorflow, Pytorch, Tensorboard
- Python libraries for QC: Qiskit, Pennylane
- Self developed libraries for EO and QC4EO: pyosv and hqm
Development Outputs
List of code and associated documentation
Project Description
[put here]