Efficiently capture and analyze aerial images with AI
How can you use artificial intelligence (AI) to get the most information from images captured by drones or satellites? Ivan Tankoyeu of AI Superior describes the process, including fundamental ideas and the main limitations of this AI solution.
As the saying goes: a picture is worth a thousand words. This expression holds true for various scenarios, from cover photos for the latest headline to personal shots of favorite moments. Images contain information—lots of it, in fact.
This is the main concept behind the use of AI for imaging. Building on the key points of our podcast episode featuring AI Superior co-founder Ivan Tankoyeu, we explore the principles underlying aerial imaging analysis and its various applications.
Steam the complete episode here.
What is AI for aerial imaging analysis?
Simply put, image analysis is a process used to better understand an image and its contents. Just like how we learned to make sense of the visual environment around us, AI is also capable of doing the same by getting information from pictures and presenting them in ways that can be useful to different fields. Some tasks that AI is now capable of doing with an image include identifying and counting objects, differentiating various elements from each other and tracking changes through time.
When used in conjunction with satellites and drones, aerial imaging analysis leads to many opportunities for visual inspections of an area. Say you want to know how land is used in a certain region. An efficient way to do this involves attaching a camera to a drone, then flying it over the selected area to take a picture. Once finished, you can inspect the resulting image for key elements.
For one picture, it seems easy and straightforward. However, the task becomes complicated if you have multiple images to analyze—each with a lot of features. It gets more challenging if you need to observe how the same area changes over time.
An AI tool built specifically for the scale of image analysis needed can greatly facilitate the whole process. After all, aerial imaging analysis can be applied in agriculture technology, environmental monitoring, forest management and property risk evaluation to name a few.
Added value that goes beyond the obvious
“In any project, it’s about the return on investment, [which] can also be estimated—not in real money or other values—but [through one’s] reputation.”
We buy certain products not only because of their utility, but also because of our trust in the brands and the added quality they bring. Consider the case of men’s watches. One would think them useless given the popularity of smartphones. However, these products still sell well. More than their use as timepieces, men’s watches are perceived to add character and refinement to the wearer.
For at least a decade, Saudi Arabia has been positioning itself as a global leader through innovative, mega city projects. AI Superior’s smart city project with a municipality in Saudi Arabia involves monitoring of trash and infrastructure, as well as analysis of recreational zones. Data from the drone images can inform key decision makers about the issues of the city that need to be prioritized: where to concentrate waste management efforts or optimize the upkeep of its recreational areas, to name a few.
Of course, one could go the usual route and employ people to do physical surveys of the area. But, this would be too inefficient, especially for tasks that require frequent monitoring. More importantly, it is a smart city project. Automation, efficiency and optimization are expected to be integrated in as many aspects of city living as possible.
When the sky's the limit for AI
“On the regulation side, It’s hard to apply this technology if you don’t have the access to collect the data.”
One obstacle to the implementation of an aerial image analysis AI tool for a project in the European Union is the General Data Protection Regulation (GDPR). It contains the set of rules and regulations regarding personal data privacy and security in the EU.
To work around this problem, satellite images can be used instead. These are available from providers like Airbus in France. However, the resolution is much smaller than what can be achieved with drone images.
From a technological perspective, data acquisition and processing for drone imaging can be made GDPR-compliant. However, this AI application still cannot be fully realized in territories like the EU where there are strong rules regarding data privacy.
