This week’s list of data news highlights covers June 12, 2021 – June 18, 2021 and includes articles about spotting wildfires and analyzing tire data about delivery vehicles.
Researchers at NASA have partnered with Khaled bin Sultan Living Oceans Foundation, an organization dedicated to ocean conservation, to map the world’s coral reefs. The project combines three types of data: the foundation’s data on coral reefs; data from NASA’s FluidCam, a remote sensing instrument that uses drones to survey underwater coral reefs; and pictures from the NeMO-Net video game, an app where citizen-scientists identify corals from 3D imagery of reefs. Researchers have already mapped about one-fifth of the world’s coral reefs.
Zillow, an online real estate company based in the United States, has created a new AI system designed to more accurately determine the value of homes across the United States. The company previously used nearly 1,000 location-specific algorithms to calculate the price of a home but now will use a single neural network to estimate prices nationwide. Zillow executives claim the new AI system will reduce price estimate errors by 11.5 percent.
Goodyear Tire & Rubber Company, a tire manufacturer based in the United States, has developed new tire features for vehicles delivering packages from e-commerce sites. The tires have sensors that collect data including tire wear, pressure, and road-surface conditions Then, the sensors transmit the data to a machine learning model, which analyzes and reports issues to vehicle operators in real time. In a pilot test, the tires accurately predicted 90 percent of issues.
An autonomous ship has departed on a journey to retrace the Mayflower voyage in 1620 and conduct oceanic research. The ship, also called the Mayflower, contains AI-powered cameras and onboard sensors to collect data on ocean acidification, microplastics, and marine life. The ship is set to arrive in Provincetown, Massachusetts, and be the largest autonomous vessel to successfully cross the Atlantic.
Researchers at the University of Hawaii have developed a model that can forecast the size and intensity of ocean waves using a database of crowdsourced photographs. The team collected photographs from local residents that showed examples of flooding of the West Maui shoreline. They used these images to train the model to find spots where coastal flooding would be particularly hazardous. Researchers can use the model to predict coastal conditions, flooding, and sea levels up to six days in advance.
Officials in Sonoma County, California have deployed an AI system to detect wildfires before they have a chance to spread. The system looks at pictures of terrain taken by 21 tower-mounted cameras and compares the images to historical photographs of the same spots. If the system detects a difference between two photographs, it alerts the county’s dispatch center. So far, the AI system has beaten human-made 911 calls by as much as 10 minutes.
Researchers at Harvard University and the Massachusetts Institute of Technology have developed in-hive sensors to monitor bee populations. Local beekeepers are using these sensors to collect data about the hive, including temperature, humidity, and bee acceleration. Researchers can then use the data to determine the viability of hives and the health of the bee population.
Researchers from the Mayo Clinic and a global volunteer consortium have created an AI system that can quickly and reliably rule out a COVID-19 infection from electrocardiography (EKG) results. The team trained the system on EKG data from 26,000 patients who were tested for COVID-19. The model predicted a negative test result with 91 percent accuracy.
Officials in the municipality of Athens, Greece have partnered with Microsoft and ATCOM, an IT services company, to deploy sensors that measure street and air pollution, pedestrian traffic, noise pollution, temperature, and humidity across the city. City officials can use the collected data to adapt city operations and resolve resident issues.
Researchers at Facebook and Michigan State University have developed an AI system that can detect deepfakes from analysis of a single image. The AI system analyzes the image’s digital fingerprints, which are unique patterns that vary based on the model used to generate the image, to determine if an image is fake and predict which model produced the image.
Image credit: Jim Maragos/U.S. Fish and Wildlife Service