Slavery from Space
High resolution satellite data were used to make a credible estimate of the number of brick kilns across the ‘Brick Belt’, helping to calculate the scale of modern slavery present. Brick kilns are high slavery-prevalent industries and before this work, the full scale of brick kilns and by proxy, slavery, was unknown, making action from the appropriate agencies difficult. This innovation provides data to help NGOs and governments fight modern slavery. This approach scales in time and space.
There is a global political commitment to ending slavery (UN Sustainable Development Goal 8.7), however, the difficulty of accurately estimating the number of slaves and their locations is a significant barrier to successful antislavery action. Remoteness in particular can make it difficult to locate and monitor sites of contemporary slavery on the ground, whilst the dynamic nature of the human system is an added challenge. Whilst a free worker cannot be distinguished from an enslaved worker from satellite remote sensing data, reliable, timely, spatially explicit and scalable metrics extracted from that satellite data can be used to reveal the location of sites associated with illegal, potentially slave-based labour. As shown in a recent documentary, NGOs on the ground in the “Brick Belt” have used the expertise from the University of Nottingham's Dr. Doreen Boyd's team to underpin rescue efforts. In short, NGOs gained more intelligence, and enslaved people were freed as a result of the data shared with NGOs by Boyd and team.
Slavery from Space specifically embraces the advantages of all forms of satellite remote sensing data (and other complementary and synergistic (geospatial) data) to unlock valuable knowledge about slavery activity and accelerate progress to its complete abolition. By combining image processing with machine learning algorithms within a number of methodological approaches, Slavery from Space aims to detect slavery but also to monitor antislavery intervention and ultimately, by being eyes in the sky, prevent slavery. The approach also includes a citizen-science component to bridge any gaps in the machine-generated knowledge, as well as aid algorithm development. A new citizen-science platform will allow citizen scientists to upload qualitative observations (e.g. notes, photographs) to further inform on spatial patterns and prevalence of slavery. Thus the objective of this innovation was to establish a methodology for using satellite data to develop reliable estimates of high slavery prevalent industries, such as brick kilns, as a proxy for slavery numbers. In common with other initiatives (e.g., the Global Partnership for Sustainable Development Data), we aim to fully harness the data revolution for antislavery and use the resultant new knowledge to eradicate slavery once and for all. Ending slavery will mean a better world for everyone: safer, greener, more prosperous, more equal. The aim is to demonstrate the role that satellite data will play in achieving this “Freedom Dividend”.
Boyd and team are currently working with private sector data providers to access even more spatially- and temporally-rich satellite data and to start thinking about other high slavery-prevalent industries to locate. The team is in dialogue with the UN to understand how to apply geospatial innovations to the wider modern slavery agenda, and open source information about the approach will be shared internationally through the UN’s Delta 8.7 knowledge platform.
Of the ILO estimates of slavery, Boyd and team estimate that a third of slavery may be detectible from space, as it takes place in stone quarries, brick kilns, fisheries, mines, forests and construction sites (rather than in domestic service, food and hospitality services, or sexual exploitation). Using satellite technology along with artificial intelligence and geospatial science methods and tools (e.g., crowd sourcing (both with proprietary and open source platforms); convoluted neural network machine learning), we can provide up-to-date, spatially explicit and defensible estimates of slavery. For example, charcoal camps are visible in Brazil, brick kilns in Cambodia, stone quarries in India, and gold mines in Tanzania. Our Slavery from Space work will extend the reach of antislavery enforcement. As one NGO (Free the Slaves) noted in a newspaper article: “Slavery from Space is a necessary addition to the Global Slavery Index, which focuses on the presence of slavery at the macro level. Slavery from Space, on the other hand, works at micro level, on the ground, and allows NGOs to tackle specific and localized cases of modern slavery. Most companies that operate illegally remain under the radar but are exposed by Slavery from Space.” The full technical methodology of one method (statistical inference and crowd sourcing) is described in ‘Slavery from Space: Demonstrating the role for satellite remote sensing to inform evidence-based action related to UN SDG 8’ in ISPRS Journal of Photogrammetry and Remote Sensing 142 (2018).
What Makes Your Project Innovative?
This innovation deploys an approach typically used in the study of fragile ecosystems to explore instead the traces that humans leave on the surface of the Earth. In doing so, it provides for the first time an accurate estimate of the prevalence within a specified geographical area of types of industry known to have high incidence of slavery. Key is that this innovative turn on data and methods can be scaled over space and time. It provides supporting evidence for anti-slavery action never seen before.
By adopting methodologies designed with other applications in mind for use with satellite data in partnership with satellite providers we have innovated how anti-slavery action is tackled. We strive to do this in as cheap and accessible way possible in order to support the drive for a slavery-free world.
What is the current status of your innovation?
Having undertaken a first round of data processing, machine learning and geospatial analysis, the satellite data is now being re-used with tweaks to methodologies to reduce the margin of error in the mapping. For instance the neural network is being retrained with more examples of kilns across a wider geographical area. We now know the limitations of the crowd contributing to the intelligence on kiln locations. Once we have eliminated any error our plan is to apply our satellite data-based approach to anti-slavery to new industries, and new locations where ground intelligence points to risk of slavery. The information and intelligence provided by the mapping will be used to inform antislavery activity in two principal ways: (i) to inform liberation and (ii) to inform the growing interest in the environmental degradation/slavery nexus which has realised that ending slavery has benefits beyond that of freeing those enslaved.
Collaborations & Partnerships
Citizens were involved as scientists through their role in visually verifying satellite images via the citizen science platform, Tomnod and Zooniverse.
Companies (e.g., DigitalGlobe, Planet) are involved through the provision of satellite data as an ‘in kind’; contribution to the innovation
Civil Society organisations (i.e., NGOs) were involved, who used the data provided by Boyd and her team to underpin rescue activity.
Users, Stakeholders & Beneficiaries
Citizens were empowered with new skills and gained personal satisfaction from using these skills in the service of antislavery.
Civil Society organisations gained new capability in understanding geospatial data, and access to accessible data and analysis about a wider geographical area than had previously been available.
Companies fulfil corporate social responsibility requirements by supporting this activity.
Results, Outcomes & Impacts
The results so far show that the new data helps NGOs plan and execute rescue missions on the ground. The results are measured on two vectors: area of territory effectively mapped (1.5ma square km); utility of resultant data for NGOs and Government (measured qualitatively and evident in the documentary file included as an attachment to this application.
It is feasible that in the future, the approach can be applied on request to support government agencies and NGOs to plan and design their missions. We are in conversation with the Office for the Special Rapporteur on Contemporary Forms of Slavery to discuss application to forthcoming missions to Italy and Togo.
It is also possible that future applications of the technology focused on understanding patterns of migration might be able to anticipate population movement, and feed data to NGOs who can help people to move safely, and thereby reduce trafficking risks and prevalence.
Challenges and Failures
When we first started working with citizen science platforms, we experimented with paid or unpaid citizen scientists analysing visual data. We had assumed that paying members of the public would increase uptake and accuracy. We actually found the opposite: unpaid citizen scientists process more images and produce more accurate data. This has informed how we will work with citizen science platforms- a crucial part of our methodology and approach – in the future.
Costs of up-to-date high resolution satellite have been something to consider. To overcome this we have established partnerships with key providers. We also are exploring the utility of free satellite data (e.g., the ESA Sentinel-2 constellation) so that future mapping is not hindered by costs and is therefore accessible to all invested in anti-slavery action.
Conditions for Success
This relies on a fluid and supportive R&D relationship between the public and private sector – the latter to provide geospatial images, the former to undertake the analysis. It requires a supportive senior management at the University, who have funded and supported the early-stage research of Dr Boyd’s team through their sponsorship of the Rights Lab (www.nottingham.ac.uk./rights-lab) to ensure that she was able to build the relationships and approaches successfully. And it has benefited from input and advice from some of the world’s leading antislavery scholars who have been able to help Dr Boyd’s team understand where to deploy their technology, and what kind of data would be valuable to NGOs.
‘Slavery from Space’ is at the stage of having been deployed ‘deeply’ in one context and is currently being evaluated with a view understanding how it can be applied in other ways, other geographies, for other industry types. We believe this innovation can be deployed by larger organisations and governments who are interested in using geospatial data to trace slavery-prevalent industry and activity.
There is much to gain by thinking laterally about data and technology and in bringing together those invested in anti-slavery action with whatever expertise that have. Having the United Nations Sustainable Development Goal highlighting the prevalence of slavery has focused action.
We are at the start of a journey that has the end goal of ending slavery by 2030. Data in all its forms is so important and this innovation is an example of Data4Good. We need support to ensure that the 40 million + people we estimate (via the GSI - https://www.globalslaveryindex.org/2018/findings/highlights/) in enslavement and would be so grateful to see this innovative use of satellite data recognised.