Our new Heat Resilience tool makes use of AI-enhanced satellite tv for pc and aerial imagery to assist cities handle dangerously excessive temperatures (and hopefully assist cut back associated well being results). The instrument combines AI-powered object detection with different fashions to investigate elements like how a lot inexperienced area a metropolis has or how effectively the roofs on buildings replicate daylight. This helps city planners and native governments determine which cooling interventions, like planting bushes, can impression temperatures most, all the best way all the way down to a neighborhood degree. To begin, we’re piloting the instrument in 14 U.S. cities, the place officers are utilizing it to find out which neighborhoods are most vulnerable to excessive warmth and develop a plan to assist. As temperatures rise globally, particularly for these dwelling in city warmth islands, AI will be a part of the answer.
Google Research has partnered with leaders within the hearth neighborhood to construct FireSat, a constellation of satellites designed particularly to detect and observe wildfires as small as a classroom (about 5×5 meters). With FireSat, authorities will be capable of entry high-resolution imagery that’s up to date globally each 20 minutes, serving to them reply to fires as rapidly as doable. It will handle a major drawback in firefighting: Till now, firefighters have had to make use of satellite tv for pc imagery that’s both low decision or solely up to date just a few occasions a day, making it onerous to seek out fires till they’ve grown bigger than a soccer area. As soon as launched, emergency responders will obtain this vital data when fires are probably simpler to comprise.
We made our bioacoustic foundation model Health Acoustic Representations, or HeAR, obtainable to researchers. HeAR is designed to assist researchers construct fashions that may hearken to human sounds, like coughs, and flag early indicators of illness. This know-how has the potential to assist display for circumstances like tuberculosis. TB is treatable, however yearly hundreds of thousands of instances go undiagnosed — actually because individuals don’t have handy entry to healthcare companies, particularly in nations with out ample medical sources. This sort of AI could assist. For example, an India-based respiratory healthcare firm is exploring how HeAR may enhance their early detection of TB based mostly on cough sounds.
