logo
AI and drones still need help from humans to find missing flood victims

AI and drones still need help from humans to find missing flood victims

Fast Company18-07-2025
For search and rescue, AI is not more accurate than humans, but it is far faster.
Recent successes in applying computer vision and machine learning to drone imagery for rapidly determining building and road damage after hurricanes or shifting wildfire lines suggest that artificial intelligence could be valuable in searching for missing persons after a flood.
Machine learning systems typically take less than one second to scan a high-resolution image from a drone, versus one to three minutes for a person. Plus, drones often produce more imagery to view than is humanly possible during the critical first hours of a search, when survivors may still be alive.
Unfortunately, today's AI systems are not up to the task.
We are robotics researchers who study the use of drones in disasters. Our experiences searching for victims of flooding and numerous other events show that current implementations of AI fall short.
However, the technology can play a role in searching for flood victims. The key is AI-human collaboration.
AI's potential
Searching for flood victims is a type of wilderness search and rescue that presents unique challenges. The goal for machine learning scientists is to rank which images have signs of victims and to indicate where in those images search-and-rescue personnel should focus. If the responder sees signs of a victim, they pass the GPS location in the image to search teams in the field to check.
The ranking is done by a classifier, which is an algorithm that learns to identify similar instances of objects—cats, cars, trees—from training data in order to recognize those objects in new images. For example, in a search-and-rescue context, a classifier would spot instances of human activity, such as garbage or backpacks, to pass on to wilderness search-and-rescue teams, or even identify the missing person themselves.
A classifier is needed because of the sheer volume of imagery that drones can produce. For example, a single 20-minute flight can produce over 800 high-resolution images. If there are 10 flights—a small number—there would be over 8,000 images. If a responder spends only 10 seconds looking at each image, it would take over 22 hours of effort. Even if the task is divided among a group of 'squinters,' humans tend to miss areas of images and show cognitive fatigue.
The ideal solution is an AI system that scans the entire image, prioritizes images that have the strongest signs of victims, and highlights the area of the image for a responder to inspect. It could also decide whether the location should be flagged for special attention by search-and-rescue crews.
Where AI falls short
While this seems to be a perfect opportunity for computer vision and machine learning, modern systems have a high error rate. If the system is programmed to overestimate the number of candidate locations in hopes of not missing any victims, it will likely produce too many false candidates. That would mean overloading squinters or, worse, the search-and-rescue teams, which would have to navigate through debris and muck to check the candidate locations.
Developing computer vision and machine learning systems for finding flood victims is difficult for three reasons.
One is that while existing computer vision systems are certainly capable of identifying people visible in aerial imagery, the visual indicators of a flood victim are often very different compared with those for a lost hiker or fugitive. Flood victims are often obscured, camouflaged, entangled in debris, or submerged in water. These visual challenges increase the possibility that existing classifiers will miss victims.
Second, machine learning requires training data, but there are no datasets of aerial imagery where humans are tangled in debris, covered in mud, and not in normal postures. This lack also increases the possibility of errors in classification.
Third, many of the drone images often captured by searchers are oblique views, rather than looking straight down. This means the GPS location of a candidate area is not the same as the GPS location of the drone. It is possible to compute the GPS location if the drone's altitude and camera angle are known, but unfortunately, those attributes rarely are. The imprecise GPS location means teams have to spend extra time searching.
How AI can help
Fortunately, with humans and AI working together, search-and-rescue teams can successfully use existing systems to help narrow down and prioritize imagery for further inspection.
In the case of flooding, human remains may be tangled among vegetation and debris. Therefore, a system could identify clumps of debris big enough to contain remains. A common search strategy is to identify the GPS locations of where flotsam has gathered, because victims may be part of these same deposits.
An AI classifier could find debris commonly associated with remains, such as artificial colors and construction debris with straight lines or 90-degree corners. Responders find these signs as they systematically walk the riverbanks and flood plains, but a classifier could help prioritize areas in the first few hours and days, when there may be survivors, and later could confirm that teams didn't miss any areas of interest as they navigated the difficult landscape on foot.
Orange background

Try Our AI Features

Explore what Daily8 AI can do for you:

Comments

No comments yet...

Related Articles

Samsung Chip Profit Tumbles While Investors Seek Bottom
Samsung Chip Profit Tumbles While Investors Seek Bottom

Yahoo

time18 minutes ago

  • Yahoo

Samsung Chip Profit Tumbles While Investors Seek Bottom

(Bloomberg) -- Samsung Electronics Co.'s chip unit plunged a far bigger-than-expected 94% in the June quarter, reflecting the depth of a slump that's plaguing the world's largest memory chipmaker. The World's Data Center Capital Has Residents Surrounded An Abandoned Art-Deco Landmark in Buffalo Awaits Revival Budapest's Most Historic Site Gets a Controversial Rebuild San Francisco in Talks With Vanderbilt for Downtown Campus We Should All Be Biking Along the Beach The weak results at the pivotal business underline how far Samsung has fallen in the lucrative artificial intelligence arena, where SK Hynix Inc. leads in high-end memory and Taiwan Semiconductor Manufacturing Co. dominates in contract chipmaking. The chip arm reported three-month operating profit of 400 billion won ($288 million), the lowest in six quarters, versus analysts' average projection for 2.73 trillion won. The decline stemmed largely from a one-time inventory cost as US export controls led to unsold AI chips at its foundry business, which relies in part on Chinese demand. The profit decline was despite solid demand for high-end memory products for servers, the company said. Shares of South Korea's largest company fell as much as 1.8% Thursday morning in Seoul. The grim results reflect Samsung's poor performance on high-bandwidth memory, said MS Hwang, research director at Counterpoint. Samsung's share in the global HBM market slipped to 17% in the second quarter in terms of bit volume, down from more than 40% in the first quarter of 2024, he said. 'This suggests that Samsung has fallen behind even Micron Technology Inc. to rank No. 3,' in the second quarter, he said. Samsung will 'actively respond' to demand for higher-density memory, the company said. Operating losses at Samsung's foundry unit are expected to narrow in the second half of the year on a gradual recovery in demand, the company said. Samsung's underwhelming quarterly report comes after the company won a $16.5 billion contract from Tesla Inc. to produce AI chips at an upcoming plant in Taylor, Texas. Its stock gained ground after news about the agreement broke earlier this week, bringing Samsung's gains in July to over 20% and putting the stock on track for its best month in years. The company has been stepping up spending on research and development and front-end capacity in its efforts to catch up with SK Hynix and Micron in AI memory chips. Samsung has struggled to get its latest products certified by Nvidia — providing an unusually long window for SK Hynix to carve out commanding leads in the booming AI memory market. At the same time, it's been trying to win large customers away from contract chipmaking powerhouse TSMC to revive its ailing foundry division where operating rates have plunged. TSMC has begun production in Arizona and is ramping up capacity in the US, while Samsung has slowed completion of its Taylor plant, now scheduled to start operating in 2026. A successful implementation of the multi-year deal with Tesla would improve Samsung's prospects for winning more clients and validate its technology for 2-nanometer mass production. Investors are also looking for clues as to whether Samsung will benefit from Nvidia Corp.'s resumption of sales of its H20 AI chips to China. The Korean memory maker's less advanced HBM3 has been used alongside H20 chips in the past. Samsung, which flagged weak preliminary operating profit and revenue numbers earlier in July, said net income came to 4.93 trillion won, missing the analysts' estimate of 6.37 trillion won. Fears about missing first-mover advantages in AI have weighed on Samsung. That's despite solid double-digit profitability in its mobile segment, which reported solid sales of its Galaxy S25 phones. Samsung hopes to maintain momentum through the recently released Galaxy Z Fold and Z Flip as well as the debut of its first trifold smartphone later this year. --With assistance from Mark Anderson. (Updates with market reaction and analyst commentary.) Russia Builds a New Web Around Kremlin's Handpicked Super App Burning Man Is Burning Through Cash It's Not Just Tokyo and Kyoto: Tourists Descend on Rural Japan Everyone Loves to Hate Wind Power. Scotland Found a Way to Make It Pay Off Cage-Free Eggs Are Booming in the US, Despite Cost and Trump's Efforts ©2025 Bloomberg L.P. Error in retrieving data Sign in to access your portfolio Error in retrieving data Error in retrieving data Error in retrieving data Error in retrieving data

Samsung profit tumbles as it struggles to catch up in the AI chip race
Samsung profit tumbles as it struggles to catch up in the AI chip race

Yahoo

time18 minutes ago

  • Yahoo

Samsung profit tumbles as it struggles to catch up in the AI chip race

Just a few years ago, Samsung Electronics stood as a leading force in the global semiconductor race, dominating one chip sector and eyeing Taiwan's TSMC in the battle for supremacy in another. But recent stumbles – including its failure to capitalize early on the artificial intelligence boom – has left it outmaneuvered, experts say, as the company struggles with deepening market share losses. Samsung's second-quarter operating profit plummeted 55% to 4.7 trillion won ($3.4 billion), down from 10.4 trillion won ($7.5 billion) a year earlier, though its revenue increased slightly compared to the same period last year. Operating profit for its chip division, historically a cash cow that used to account for two-thirds of its total profit, shrank by nearly 94% from April to June compared to a year ago. It blamed the worse-than-expected performance on inventory value adjustments, low utilization rate for its contract chipmaking business and continued fallout from US export controls on advanced AI chips to China – a key market for Samsung. Thursday's disappointing earnings report reignites concerns about the future of the embattled South Korean tech giant. Samsung warned investors of its dismal performance in its earnings projection earlier this month. The results come on the heels of a $16.5 billion deal with Tesla, announced this week, to produce its new chips – a move expected to boost Samsung's outlook. Looking ahead for the second half the year, Samsung said it plans to proactively meet the growing demand for high value-added and AI-driven products and continue to strengthen competitiveness in advanced semiconductors. How Samsung lost its edge South Korea's largest conglomerate has run into significant headwinds in recent years across both of its key revenue streams: the manufacturing of memory chips, which help devices store data, and logic chips, which power data processing and computation. Once the industry's leading memory chipmaker, Samsung has lost ground to rivals like South Korean SK Hynix and American Micron Technology, particularly in the fast-growing market for high bandwidth memory (HBM) sector. HBM, made up of stacks of DRAM memory chips (dynamic random access memory) used for short-term data storage, are essential for AI processors developed by companies like Nvidia and AMD. Meanwhile, Samsung's logic semiconductor business trails industry leader TSMC both in cutting-edge chip technologies and market share. In the first quarter of this year, SK Hynix overtook Samsung to lead the global DRAM market, while TSMC extended its dominance in logic chips with a 68% market share, compared to just 8% for Samsung, according to market research firm TrendForce. Sanjeev Rana, head of Korea research at CLSA, a brokerage firm, said a series of 'missteps' by Samsung – most notably management's failure to anticipate the surge in AI demand – has contributed to its current struggles. 'They were slow to recognize the coming AI revolution, and they bet on some other products, other technologies, which, in hindsight, didn't turn out to be very good bets,' he said, explaining that Samsung overlooked the potential of HBM initially. As a result, Samsung has so far missed out on being a supplier for its most advanced high-bandwidth memory product to Nvidia, which accounts for nearly 80% of global HBM demand last year, according to Rana. The product has repeatedly failed Nvidia's performance tests, though he expects the company to clear them in the next two months. While Samsung announced in June that it managed to secure orders from AMD and Broadcom, rivals SK Hynix and Micron have already begun delivering samples of more advanced memory chips to customers. At the same time, Samsung's logic chip business – once central to its ambition to rival TSMC – is also under mounting pressure. Despite tens of billions of investments over the past few years, the company has been unable to secure meaningful orders for its advanced chips, leading to underutilized facilities, Rana said. Last year, CLSA estimated that Samsung's contract chipmaking business posted an operating loss of 5.6 trillion won ($4.1 billion). That figure is expected to rise to 6.6 trillion won ($4.8 billion) this year. US restrictions on the sale of advanced chips to China have also taken a toll on Samsung's revenue, as shipments to Chinese clients and projects were forced to pause pending regulatory review, said Joanne Chiao, an analyst at TrendForce. But with some chips now having cleared the review process, the second quarter is expected to be the most affected period, she added. A potential turnaround thanks to Tesla Tesla offered Samsung a lifeline this week. Its CEO Elon Musk announced that the electric vehicle company has tapped the Korean chipmaker to make its new chips for self-driving cars and humanoid robots in a $16.5 billion deal. 'Samsung's giant new Texas fab will be dedicated to making Tesla's next-generation AI6 chip,' he said in a post on X. 'Samsung agreed to allow Tesla to assist in maximizing manufacturing efficiency. This is a critical point, as I will walk the line personally to accelerate the pace of progress.' Samsung's shares surged more than 6.9% to reach their highest level since September following news of the deal. Tesla currently sources its AI4 chips, which power its advanced driver assistance systems called Full Self-Driving (FSD) software, from Samsung, but it enlisted TSMC to produce its AI5 chips, according to Musk. The deal came after Samsung postponed the operational start of its chipmaking plants in Taylor, Texas to 2026 from its original schedule of 2024, as it struggled to win customers for the project. Ray Wang, research director focusing on semiconductor industry at Futurum Group, called the deal with Tesla 'significant,' saying it could boost Samsung's struggling profitability and validate its capabilities in producing advanced chips. The agreement will also help increase utilization of its Texas facilities, improving the company's return on investment, he added. Rana said that although mass production for the Tesla project won't begin until 2027, the deal is a boost to market sentiment and represents 'a big word of confidence.' 'The management has done a lot of restructuring for this business in the last 12-15 months or so, so I think they now understand what the problems were, and they have made some efforts to resolving those issues,' he said. 'Things will get better from the second half (of the year).'

DOWNLOAD THE APP

Get Started Now: Download the App

Ready to dive into a world of global content with local flavor? Download Daily8 app today from your preferred app store and start exploring.
app-storeplay-store