How AI is enhancing aerial imagery to help identify items usually recognized by humans

With the help of AI, organizations are finding ways to turbocharge photographic imagery so that it comes with both content and context.

A construction company can now gain access to aerial imagery of an entire project and map this imagery to what its project lists in terms of tasks completed and those underway or waiting. There no longer is a constant need to physically travel to a site for a physical inspection.

SEE: Artificial Intelligence Ethics Policy (TechRepublic Premium)

“You can use artificial intelligence to turn high-resolution aerial imagery into a powerful project management tool for a variety of industries, including construction, insurance and government agencies,” said Tony Agresta, general manager of North America for Nearmap, which provides aerial imagery and geospatial data. “Using AI, these organizations can virtually inspect, measure and analyze hundreds of location insights such as property lines, shrubbery on a property, roof and building dimensions, vegetation in areas where animals can be relocated and more. Viewing perspectives can be rotated, tilted and changed to see an image from any angle.”

This is a giant step forward from past practice, when the aerial imagery organizations relied on was often out of date, lacked data and context and was difficult to integrate with other software.

Today, systems can capture and process aerial imagery. Companies can use software to custom-label images. They can employ machine learning model architectures, data processing and tools to extract and interact with the data.

“For every aerial survey that’s conducted, the AI system is running to detect dozens of objects,” Agresta said. “By utilizing human expert labeling to train the models, machine learning models can mimic a human’s ability to recognize features in an image.”

Agresta points to how the Australian Red Cross used advanced visual imagery and AI during the 2019-2022 brushfires to verify claims for assistance.

“Initially offering grants of up to $70,000 to those hit the hardest, the Australian Red Cross was looking for a way to streamline how funds could be allocated to people in need,” Agresta said. ”By using visual imagery supplemented with AI, the Australian Red Cross was able to access high-resolution aerial imagery that was captured within days following the fires to visually assess whether the damage [from brush fires] was a factor present in incomplete applications for aid it had received. This gave them the information they needed to keep grants moving along with ease.“

The ability to solve real-world problems with aerial imagery and software is enhanced by systems that now update these images more often so images stay current. Aerial imagery systems also come with APIs (application programming interfaces) that allow them to be used with a wide range of systems and applications.

SEE: How aerial photography integrated with GIS systems can help make cities smarter (TechRepublic)

Aerial imagery is a great option for those looking to save time and money because it makes it possible for site visits and inspections to be done virtually,” Agresta said. “But If you are looking to make more decisions based on aerial imagery, there are a few key considerations, such as making sure that the aerial imagery provider you select can give you the insights and information necessary for your project.”

Important points you will want to check on with any aerial imagery provider are how accurate the depth and height measurements or the identification of a particular detail are, since the AI operating on the data will only be as good as the data it works on.

Additionally, you should also check the compatibility between the systems you’re running in-house or on the cloud and the aerial imagery you want to use with them.

“There is also one other consideration,” Agresta said. “You should always compare current images with historical images to make sure there is nothing you’ve accidentally overlooked when it comes to seasonal changes or climate change.”