14:00–14:05
Pier Luigi Dovesi · The Good AI Lab
CV4GOODComputer Visionfor Humanitarian Action
1st
edition
edition
ECCV 2026 workshop
Tuesday, September 8, 2026
14:00–18:00 CEST · Malmö, Sweden
Malmömässan · Room D2
CV4GOOD is an ECCV 2026 workshop bringing computer vision research face to face with frontline humanitarian practice in Malmö, Sweden. Born from a partnership between The Good AI Lab and Médecins Sans Frontières, it pairs researchers building vision and Earth-observation pipelines with practitioners who can say what works in the field, what fails, and what is missing.
co-hosted by
The Good AI Lab× Médecins Sans Frontières
A workshop shaped jointly by frontline humanitarian practice and computer-vision research.
speakers & moderator
Whoyou’llhear from
The people grounding the workshop in frontline practice, deployed methods, and governance.
field · research · governance
The program mixes humanitarian practice, academia, and industry research. It opens with needs from the field and closes with a panel on dual use and governance.
program
Half-DayWorkshopProgram
The workshop takes place in person in Malmö and is streamed to registered ECCV 2026 attendees. The first half grounds the audience in humanitarian needs and deployed systems; the second broadens to methods, governance, and open discussion.
Attendance is included with ECCV 2026 registration.
Malmömässan · Room D2
Tuesday 8 September 2026
14:00–18:00 CEST · in person + streamed
All times are local to Malmö (CEST). Talk slots include Q&A.
14:05–14:30
Yann Rebois · MSF
Needs from the field
What MSF uses, what breaks, and what the research community should build next.
14:30–14:50
Lorenz Wendt · GEOHUM
Earth observation for humanitarian action
Where satellite EO and GIS meet humanitarian needs — navigating users, funding, and research.
14:50–15:10
Jamon Van Den Hoek · Conflict Ecology Lab
Conflict monitoring from orbit
Satellite imagery, radar-based damage assessment, and near-real-time monitoring in active conflict.
15:10–15:30
Corey Scher · Conflict Ecology Lab
Conflict monitoring from orbit
InSAR damage assessment that sees through clouds and works at night — near-real-time, city-scale mapping under pressure.
15:30–16:00
Coffee break
16:00–16:30
Sara Beery · MIT
Targeting human verification for robust AI in ecology
Where to spend scarce expert review when the model meets a new domain.
16:30–17:00
Joseph Redmon · OlmoEarth, Ai2
Open foundation models for earth observation
From dual-use concerns to open EO infrastructure for non-profits and NGOs.
17:00–17:55
Closing Panel · Moderated by Federico Tombari
Dual-use, governance, and the research agenda
All invited speakers. Open discussion bridging humanitarian needs, computer-vision methods, and responsible deployment.
17:55–18:00
Closing remarks
workshop
Topics ofinterest
CV4GOOD is a half-day ECCV workshop that puts computer-vision researchers in the same room as the people who actually use these tools in humanitarian settings.
We want field constraints to shape the research agenda, and collaborations that outlast the session.
why this matters
Computer vision already feeds humanitarian decisions. What is mostly missing is evidence that these systems hold up outside the conditions they were built in, and accountability when they do not.
co-hosted by
The Good AI Lab ×
Médecins Sans Frontières
Humanitarian AI and computer vision
Earth observation and remote sensing
Conflict, displacement, and food insecurity
Responsible deployment in high-stakes settings
Satellite damage assessment
Optical and SAR pipelines for conflict and disaster response.
Refugee settlement mapping
Detection and monitoring of displacement camps from space.
Crop stress and food security
Multispectral and temporal sensing for early warning under pressure.
Foundation models for earth observation
Transferable representations, few-shot adaptation, and multimodal reasoning.
Trustworthy AI for lives and aid
Calibration, auditability, and failure modes when a wrong output costs more than a benchmark point.
deployment constraints
Domain shift across geographies and time
Scarce and noisy labels
Auditability
Calibrated uncertainty
Integration into GIS and field workflows
organizing committee
The peoplebehindCV4GOOD
CV4GOOD brings together researchers, practitioners, and organizers across academia, humanitarian work, and applied AI to shape a field-informed workshop from the ground up. These are the people making it happen.
in partnership with
Pier Luigi Dovesi
The Good AI Lab · AMD Silo AI
Matteo Poggi
University of Bologna
Federico Tombari
The Good AI Lab · Google · TU Munich
Philip Torr
University of Oxford
Ann De Schutter
Médecins Sans Frontières
Giorgos Tolias
The Good AI Lab · Czech Technical University in Prague
Noa Garcia
The University of Osaka
Bill Psomas
The Good AI Lab · Czech Technical University in Prague
Gianluca Villani
The Good AI Lab
Theodoros Panagiotakopoulos
The Good AI Lab · King
Rafailia Vlachou
The Good AI Lab · SEB
Maria Rosaria Briglia
The Good AI Lab · Sapienza University of Rome
Mattia Segù
The Good AI Lab · Google
Carlo Rapisarda
The Good AI Lab · GeoGuessr
The Good AI LabSocial
Keep the conversation going after CV4GOOD with workshop attendees, speakers, The Good AI Lab community, and people from across AI research and industry. Hosted at King, part of Microsoft Gaming.