How AI can really be helpful in disaster response

Marash, Turkey: Satellite images (left) from earth imaging company Planet Labs PBC and results from xView2 (right) provided by UC Berkeley, the Defense Innovation Unit, and Microsoft.

This is an improvement over more traditional disaster assessment systems, in which emergency responders and rescuers rely on reports and witness calls to quickly identify where help is needed. In some recent cases, fixed-wing aircraft such as drones have flown over disaster areas with cameras and sensors attached to provide human-reviewed data, but this still has implications. may happen. take days, if not longer. Typical responses are even slower due to the fact that the different responding organizations often have their own data catalogs, making it possible to create a shared, standardized picture of the areas in need of help. become difficult. xView2 can create a shared map of the affected area in minutes, helping organizations coordinate and prioritize responses—saving time and lives.


Of course, this technology is not a panacea for disaster response. There are several major challenges facing xView2 that are currently attracting much of Gupta’s research attention.

First and foremost is how dependent the model is on satellite imagery, which only yields clear images during the day, when there is no cloud cover, and when the satellite is overhead. The first usable images out of Turkey did not appear until February 9, three days after the first earthquake. And there are very few satellite images taken in remote and economically underdeveloped areas — such as right across the border in Syria. To solve this problem, Gupta is working on new imaging techniques such as synthetic aperture radar, which creates images using microwave pulses instead of light waves.

Second, although the xView2 model is 85 or 90% accurate in accurately assessing damage and severity, it cannot actually detect damage on the sides of the vehicle. buildings, as satellite images have an aerial view.

In the end, Gupta says it’s very difficult for actual organizations to use and trust an AI solution. “First responders are very traditional,” he said. “When you start telling them about this fancy AI model, which isn’t even on the ground and it’s looking at pixels from 120 miles away in space, they won’t believe anything. ”

What’s next?

xView2 supports multiple phases of disaster response, from immediate mapping of damaged areas to assessment of possible safe shelter sites to long-term reconstruction coverage . Abbhi, first of all, says he hopes xView2 “will be really important in our inventory of damage assessment tools” at the World Bank in the future.

Since open source and this program is free, anyone can use it. And Gupta intends to stay that way. “When companies come in and start saying, We can commercialize this, I hate that,” he said. “This has to be a public service run for the benefit of everyone.” Gupta is working on a web application so any user can run a review; Currently, organizations have contacted xView2 researchers for analysis.

Source by [author_name]


News7h: Update the world's latest breaking news online of the day, breaking news, politics, society today, international mainstream news .Updated news 24/7: Entertainment, the World everyday world. Hot news, images, video clips that are updated quickly and reliably

Related Articles

Back to top button