Covering Disruptive Technology Powering Business in The Digital Age

Written by: Martin Dale Bolima, Tech Journalist, AOPG.

You can count NVIDIA among the growing list of tech titans looking to save the world.

NVIDIA announced in a virtual media brief that it is now working with the United Nations Satellite Centre (UNOSAT) to leverage Artificial Intelligence (AI) and Deep Learning in working toward a healthier, more sustainable Earth. This announcement, incidentally, comes a day after Oracle reaffirmed—also in a virtual media brief—its commitment to sustainability by utilising its leading data capabilities.

“Climate change is the biggest threat facing humanity today, and there is a very high cost of delayed action. We have to act now [and save the planet],” said Shilpa Kolhatkar, Head of Global AI Nations Business Development at NVIDIA.

That cost is becoming more apparent, with flooding in various countries, for instance, worsening by the year. This problem has so far affected an estimated 2.2 billion people globally and is causing nations USD $40 billion in damages every single year. Those living in coastal areas—which comprise about 40% of the world’s total population—have it worse as flooding is oftentimes exacerbated by rising sea levels worldwide and extreme weather events.

Enhancing AI for Earth Observation

Catastrophic events, like massive flooding, and similar disasters are occurring with alarming regularity and make climate-related disaster management an imperative today. That is exactly the purpose of AI for Earth Observation (AI4EO)—the umbrella term that encompasses AI-related initiatives to monitor and assess the many changes Earth is undergoing, including the unabated warming of the atmosphere that is then causing climate change.

The reason why AI for Earth Observation is crucial is for a couple of factors. Flood level estimation needs to be done remotely because the physical access to flooded areas is limited, and deploying instruments to potential flood zones can be dangerous,” explained Kolhatkar. “Couple that with the availability of high-definition, high-resolution images that are available from satellites… but the process of analysing and labelling these satellite imagery is time-consuming and is not scalable for huge amounts of data.”

NVIDIA’s AI expertise is about to change all that. Moving forward, UNOSAT’s satellite imagery technology infrastructure will be integrated with NVIDIA’s accelerated computing platform, thus enabling the fast-tracking of AI4EO efforts in research. This AI-powered, NVIDIA-enhanced system will collect and analyse geospatial information in real-time to give key decision-makers real-time insights about climate-related disasters, like flooding, super typhoons and wildfires.

Leveraging Data With the NVIDIA Deep Learning Institute

Additionally, UNOSAT is building on the NVIDIA Deep Learning Institute (DLI) course, which Kolhatkar describes as being the handiwork of NVIDIA’s data scientists in India. Their work, according to Kolhatkar, “demonstrated that what would otherwise take days of manual work to determine flooded regions in satellite images can be achieved within seconds with the use of Deep Learning and AI.”

“Our partnership with UNOSAT drives a twofold strategy. First, it gives us the unique opportunity to help advance UN’s Sustainable Development Goals to help in humanitarian aid and disaster relief and through GPU-accelerated models that improve on flood impact predictions,” said Kolhatkar in explaining the NVIDIA-UNOSAT collaboration on the DLI. “Secondly, it supports our goal of training data scientists and training climate researchers around the world on how AI can be used to combat the impact of climate change because a skilled workforce is needed to make AI and data science work.”

On top of NVIDIA’s courses, UNOSAT has also developed its own module to augment DLI’s free online course with something related to sustainability. The UNOSAT module, titled “Disaster Risk Monitoring Using Satellite Imagery,” is the first NVIDIA DLI course focused on climate action, and it specifically covers how to build a Deep Learning model to automate the detection of flood events.

According to Einar Bjørgo, Director at UNOSAT, the collaboration between NVIDIA and UNOSAT on DLI has helped the latter utilise data at scale, giving them greater confidence in the products they are releasing.

“Through the years, we have really had quite a bit of experience in developing proper flood data and statistics but we struggled to do it at scale,” said Bjørgo. “We have been able to use NVIDIA algorithms, coupled with our own data scientists. We’ve had them make sure that the different outputs are really at scale.”

The “Disaster Risk Monitoring Using Satellite Imagery” module is based on a flood in Nepal, and it is being offered for free. Its goal is to upskill data scientists around the world to harness accelerated computing to predict and respond to climate-related disasters.

“Working with NVIDIA will enable us to close the loop from AI research to implementation of climate solutions in the shortest time possible, ensuring that vulnerable populations can benefit from the technology,” Bjørgo added.

The Devil Is in the Details

Ultimately, decisions on anything are only as good as the quality of the data available.

To that end, UNOSAT will soon be able to get better images as it has supercharged its satellite imagery technology infrastructure with NVIDIA DGX systems and the NVIDIA EGX platform. The former enables AI development at scale, while the latter enables accelerated computing from the data centre to the edge. The result, according to UNOSAT, is faster, better, and more accurate AI-based flood detection that covers larger areas.

These initiatives are par for the course for NVIDIA, whose climate change-related pursuits are well documented. Its work with UNOSAT is yet another example, and it may have large-scale implications in the fight against climate change.

(0)(0)