Earning Its Stripes: Tech Used To Crack Tiger Trade
August 18th, 2022
Via Terra Daily, an article on the use of artificial intelligence to help stop poaching:
In a town in northeastern Scotland, Debbie Banks looks for clues to track down criminals as she clicks through a database of tiger skins.
There are thousands of photographs, including of rugs, carcasses and taxidermy specimens.
Banks, the crime campaign leader for the Environmental Investigation Agency (EIA), a London-based charity, tries to identify individual big cats from their stripes.
Once a tiger is identified, an investigator can pinpoint where it comes from.
“A tiger’s stripes are as unique as human fingerprints,” Banks told AFP.
“We can use the images to cross-reference against images of captive tigers that might have been farmed.”
Currently this is slow painstaking work.
But a new artificial intelligence tool, being developed by The Alan Turing Institute, a centre in the UK for data science and artificial intelligence, should make life much easier for Banks and law enforcement officials.
The project aims to develop and test AI technology that can analyse the tigers’ stripes in order to identify them.
“We have a database of images of tigers that have been offered for sale or have been seized,” Banks said.
“When our investigators get new images, we need to scan those against the database.
“At the moment we are doing that manually, looking at the individual stripe patterns of each new image that we get and cross-referencing it against the ones we have in our database.”
It is hoped that the new technology will help law enforcement agencies determine where tiger skins come from and allow them to investigate the transnational networks involved in trafficking tigers.
Once the officials know the origins of confiscated tiger skins and products, they will be able to tell whether the animal was farmed or poached from a protected area.
Poaching, fuelled by consumer demand, remains a major threat to the survival of the species, according to the EIA.
Tiger skins and body parts are sought after, partly due to their use in traditional Chinese medicine.
An estimated 4,500 tigers remain in the wild across Asia.
“Tigers faced a massive population decline in the last 120 years, so we want to do everything we can to help end the trade in their parts and products, including tiger skins,” Banks said.
Anyone with photographs of tigers is invited to submit them to the EIA to help bolster the AI database.
“We are inviting individuals — whether they are photographers or researchers and academics — who may have images of tigers where their stripe patterns are clear,” Banks said.
“They could be live tigers, dead tigers or tiger parts.
“If they can share those with us, the data scientists can then develop, train and test the algorithm,” she said.
“We need thousands of images just to do that phase of the project.”
,
Read More »Google’s Dynamic World Uses AI To Analyze Satellite Data
June 10th, 2022
Via Fast Company, an article on a new tool from Google that shows how the planet is changing in near real time:
The planet changes quickly: More than half a million acres are burning in New Mexico. A megadrought is shrinking Lake Mead. The Alps are turning from white to green. Development continues to expand, from cities to massive solar farms. All of these changes impact the Earth’s climate and biodiversity. But in the past, such changes have been difficult to track in detail as they’re happening.
A new tool from Google Earth Engine and the nonprofit World Resources Institute pulls from satellite data to build detailed maps in near real time. Called Dynamic World, it zooms in on the planet in 10-by-10-meter squares from satellite images collected every two to five days. The program uses artificial intelligence to classify each pixel based on nine categories that range from bare ground to trees, crops, and buildings.
Researchers, nonprofits, and other users can “explore and track and monitor changes in these terrestrial ecosystems over time,” says Tanya Birch, senior program manager for Google Earth Outreach. As the tool was being built last year, Birch used it in the days after the Caldor Fire, a wildfire that burned more than 200,000 acres in California. The pixels in satellite images quickly changed from being classified as “trees” to “shrub and scrub.”
Scientists used to rely on statistical tables that were sometimes released only every five years, says Fred Stolle, deputy director of the World Resources Institute’s Forests Program. “That’s clearly not good enough anymore,” he says. “We’re changing so fast, and the impact is so fast, that satellites are now the way to go.”
Researchers and planners already use satellite data in some applications—the World Resources Institute, for example, previously worked with Google to build Global Forest Watch, a tool that can track deforestation using satellite images. But the new data is much more detailed; now it’s sometimes possible to see if one or two trees are cut down in a tropical forest, even when a larger area is intact, Stolle says.
In cities, planners could use the data to easily see which neighborhoods don’t have enough green space. Researchers studying smallholder farms in Africa could use it to see the impacts of drought and when crops are being harvested. Because the data is continuously updated, it’s also possible to watch the seasons change throughout the year across the entire planet. The data goes back five years, and using the new tool, anyone can enter date ranges to see how a location has changed over time.
“I encourage people to dive into it and explore,” Birch says. “There’s a lot of depth and a lot of richness in Dynamic World. . . . I feel like this is really pushing the frontier of mapmaking powered by AI in an incredibly novel way.”
,
Read More »Satellites and AI Can Help Solve Big Problems—If Given the Chance
June 6th, 2022
Via Wired, a report on some of the hurdles that stand in the way of ambitious plans to use imagery to help feed people, reduce poverty, and protect the planet:
For the past three decades, three decades, geologist Carlos Souza has worked at the Brazil-based nonprofit Imazon, exploring ways he and the teams he coordinates can use applied science to protect the Amazon rainforest. For much of that time, satellite imagery has been a big part of his job.
In the early 2000s, Souza and colleagues came to understand that 90 percent of deforestation occurs within 5 kilometers of newly created roads. While satellites have long been able to track road expansion, the old way of doing things required people to label those findings by hand, amassing what would eventually become training data. Those years of labor paid off last fall with the release of an AI system that Imazon says reveals 13 times more roadway than the previous method, with an accuracy rate of between 70 and 90 percent.
Proponents of satellite imagery and machine learning have ambitious plans to solve big problems at scale. The technology can play a role in anti-poverty campaigns, protect the environment, help billions of people obtain street addresses, and increase crop yields in the face of intensifying climate change. A UNESCO report published this spring highlights 100 AI models with the potential to transform the world for the better. But despite recent advances in deep learning and the quality of satellite imagery, as well as the record number of satellites expected to enter orbit over the next few years, ambitious efforts to use AI to solve big problems at scale still encounter traditional hurdles, like government bureaucracy or a lack of political will or resources.
Stopping deforestation, for instance, requires more than spotting the problem from space. A Brazilian federal government program helped reduce deforestation from 2004 to 2012 by 80 percent compared to previous years, but then federal support waned. In keeping with an election promise, President Jair Bolsonaro weakened enforcement and encouraged opening the rainforest to industry and cattle ranch settlers. As a result, deforestation in the Amazon reached the highest levels seen in more than a decade.
Other AI-focused conservation groups have run into similar issues. Global Fishing Watch uses machine learning models to identify vessels that turn off GPS systems to avoid detection; they’re able to predict the type of ship, the kind of fishing gear it carries, and where it’s heading. Ideally that information helps authorities around the world target illegal fishing and inform decisions to board boats for inspection at sea, but policing large swaths of the ocean is difficult. Global Fishing Watch’s tech spotted hundreds of boats engaged in illegal squid fishing in 2020, data that head of research David Kroodsma credits with increasing cooperation between China and South Korea, but it didn’t lead to any particular prosecution. Enforcement in ports, he says, is “key to making deterrence scalable and affordable.”
Back on land, the consulting company Capgemini is working with The Nature Conservancy, a nonprofit environmental group, to track trails in the Mojave Desert and protect endangered animal habitats from human activity. In a pilot program last year, the initiative mapped trails created by off-road vehicles in hundreds of square miles of satellite imagery in Clark County, Nevada, to create an AI model that can automatically identify newly created roads. Based on that work, The Nature Conservancy intends to expand the project to monitor the entirety of the desert, which stretches more than 47,000 square miles across four US states.
However, as in the Amazon, identifying problem areas only gets you so far if there aren’t enough resources to act on those findings. The Nature Conservancy uses its AI model to inform conversations with land managers about potential threats to wildlife or biodiversity. Conservation enforcement in the Mojave Desert is overseen by the US Bureau of Land Management, which only has about 270 rangers and special agents on duty.
In northern Europe, the company Iceye got its start monitoring ice buildup in the waters near Finland with microsatellites and machine learning. But in the past two years, the company began to predict flood damage using microwave wavelength imagery that can see through clouds at any time of day. The biggest challenge now, says Iceye’s VP of analytics, Shay Strong, isn’t engineering spacecraft, data processing, or refining machine learning models that have become commonplace. It’s dealing with institutions stuck in centuries-old ways of doing things.
“We can more or less understand where things are going to happen, we can acquire imagery, we can produce an analysis. But the piece we have the biggest challenge with now is still working with insurance companies or governments,” she says.
“It’s that next step of local coordination and implementation that it takes to come up with action,” says Hamed Alemohammad, chief data scientist at the nonprofit Radiant Earth Foundation, which uses satellite imagery to tackle sustainable development goals like ending poverty and hunger. “That’s where I think the industry needs to put more emphasis and effort. It’s not just about a fancy blog post and deep learning model.”
It’s often not only about getting policymakers on board. In a 2020 analysis, a cross-section of academic, government, and industry researchers highlighted the fact that the African continent has a majority of the world’s uncultivated arable land and is expected to account for a large part of global population growth in the coming decades. Satellite imagery and machine learning could reduce reliance on food imports and turn Africa into a breadbasket for the world. But, they said, lasting change will necessitate a buildup of professional talent with technical knowledge and government support so Africans can make technology to meet the continent’s needs instead of importing solutions from elsewhere. “The path from satellite images to public policy decisions is not straightforward,” they wrote.
Labaly Toure is a coauthor of that paper and head of the geospatial department at an agricultural university in Senegal. In that capacity and as founder of Geomatica, a company providing automated satellite imagery solutions for farmers in West Africa, he’s seen satellite imagery and machine learning help decision-makers recognize how the flow of salt can impact irrigation and influence crop yields. He’s also seen it help settle questions of how long a family has been on a farm and assist with land management issues.
Sometimes free satellite images from services like NASA’s LandSat or the European Space Agency’s Sentinel program suffice, but some projects require high-resolution photos from commercial providers, and cost can present a challenge.
“If decision-makers know [the value] it can be easy, but if they don’t know, it’s not always easy,” Toure said.
Back in Brazil, in the absence of federal support, Imazon is now forging ties with more policymakers at the state level. “Right now, there’s no evidence the federal government will lead conservation or deforestation efforts in the Amazon,” says Souza. In October 2022, Imazon signed cooperation agreements with public prosecutors gathering evidence of environmental crimes in four Brazilian states on the border of the Amazon rainforest to share information that can help prioritize enforcement resources.
When you prosecute people who deforest protected lands, the damage has already been done. Now Imazon wants to use AI to stop deforestation before it happens, interweaving that road-detection model with one designed to predict which communities bordering the Amazon are at the highest risk of deforestation within the next year.
Deforestation continued at historic rates in early 2022, but Souza is hopeful that through work with nonprofit partners, Imazon can expand its deforestation AI to the other seven South American countries that touch the Amazon rainforest.
And Brazil will hold a presidential election this fall. The current leader in the polls, former president Luiz Inácio Lula da Silva, is expected to strengthen enforcement agencies weakened by Bolsonaro and to reestablish the Amazon Fund for foreign reforestation investments. Lula’s environmental plan isn’t expected out for a few months, but environmental ministers from his previous term in office predict he will make reforestation a cornerstone of his platform.
,
Read More »These New Technologies Could Transform Wildlife Conservation
March 22nd, 2022
Via The Hill, a look at how artificial intelligence, environmental DNA and networked sensors are among the technologies with the highest potential to improve wildlife conservation:
Published last December by conservation technology network WILDLABS, together with a group of non-profit and academic partners, the report is the first of its kind to provide a holistic assessment of the state of conservation technology.
The researchers surveyed 248 conservationists, technologists and academics across 37 countries over the 11 most commonly used conservation technologies, including camera traps, biologgers, acoustic monitoring and remote sensings.
Although it’s estimated that about 8.7 million species populate our planet, 86 percent of all species on land and 91 percent in the oceans are yet to be discovered. Multiple scientific studies suggest that if no action is taken, as many as half of all species could go extinct by the end of the century.
Traditional methods for tracking biodiversity, such as camera traps, which connect digital cameras to an infrared sensor to capture images and videos of animals moving past the sensor, or aerial surveys can be labor-intensive and costly. The technologies highlighted by the research could help reduce the time and resources required to detect wildlife, while increasing the effectiveness of conservation efforts.
Combining AI and citizen science to improve wildlife identification
Artificial intelligence (AI) is increasingly used to analyze large amounts of conservation data, such as camera trap, satellite and drone images or audio and video recordings, and improve wildlife identification and monitoring. The non-profit Wild Me created a cloud-based platform Wildbook, which uses computer vision and deep learning algorithms to scan millions of crowdsourced wildlife images to identify species and individual animals based on their unique patterns, including stripes, spots or other defining physical features such as scars.
Photos are added to the cloud by scientists and other volunteers, or are sourced from social media, and over time, the information about each species will grow as more citizen scientists and researchers contribute to the image catalogue. The aggregated data helps inform conservation actions, while the public can follow their favorite animals in the cloud.
Wildbook was started off to improve the tracking of whale sharks which was previously done by attaching plastic tags to the animals that had often never resurfaced. The platform has since grown into a vast database of various different species, including sea turtles, manta rays, sharks, whales, dolphins, big cats, giraffes and zebras.
In partnership with Microsoft’s AI for Earth initiative, Wildbook is hosted on its cloud computing service, Azure and is made available as an open-source software to encourage others to adopt this non-invasive method of species tracking.
A facial recognition tool for wildlife
The BearID Project is developing a facial recognition software that can be applied to camera trap imagery to identify and monitor brown bears, and inform subsequent conservation measures. This is especially important because camera traps are currently unable to consistently recognize individual bears due to the lack of unique natural markings for certain species.
So far, the team of biologists and software engineers have developed an AI system using personal photographs of brown bears from British Columbia, Canada and Katmai National Park, Alaska, which was able to recognize 132 individual bears with an 84 percent accuracy. While the camera trap system is currently under development, the project is already working with indigenous nations in Canada to implement the new tool within bear research and monitoring programs. The ultimate goal is to expand the scope of the facial recognition software to eventually apply to other threatened species.
Using AI to combat wildlife trafficking
AI can also help boost anti-poaching efforts. The software Protection Assistant for Wildlife Security (PAWS) takes in past poaching records and the geographic data of the protected area to predict poachers’ future behavior, and design poaching risk maps and optimal patrol routes for rangers.
During the first month of its field tests in the Srepok Wildlife Sanctuary in Cambodia, the area identified as most suitable for the reintroduction of tigers in Southeast Asia, PAWS has helped rangers double the amount of snares detected and removed during their patrols.
PAWS has since been integrated with the open-source Spatial Monitoring and Reporting Tool (SMART), which is already used by rangers in over 1,000 protected areas to log data collected during patrols. The integrated tool is currently available to national parks as a beta feature, and has been tested across Zimbabwe, Nigeria, Kenya, Malaysia, Mozambique and Zambia to generate poaching risk maps to assist with patrols.
Plans for the future include connecting the software to remote sensing tools such as satellites or drones to reduce the need for humans to enter the data, and expanding the scope of PAWS to predict other forms of environmental crime, including illegal logging or fishing.
Sampling environmental DNA for biodiversity monitoring
Environmental DNA (eDNA), meanwhile, enables conservationists to collect biodiversity data by extracting DNA from environmental samples, such as water, soil, snow or even air. All living organisms leave traces of their DNA in their environments through their feces, skin or hair, amongst others.
A single sample might carry the genetic code of tens or even hundreds of species, and can provide a detailed snapshot of an entire ecosystem. A recent study has revealed that eDNA could offer a more efficient and cost-effective method for the large-scale monitoring of terrestrial biodiversity. In the study, eDNA sampling detected 25 percent more terrestrial mammal species compared to camera traps, and for half of the cost.
eDNA can also help examine the impact of climate change, detect invisible threats such as viruses or bacteria, and assess the overall health of an ecosystem, which can be used to make the case for greater protection for the area.
NatureMetrics, for instance, partnered with the Lebanon Reforestation Initiative to use eDNA to assess the biodiversity of freshwater ecosystems, providing crucial data from a previously understudied region to inform rehabilitation and restoration work.
Increasing connectivity for better conservation outcomes
By enabling camera traps, tracking devices and other conservation hardware to connect online, networked sensors can offer a more comprehensive picture of animal behavior and provide instant alerts about imminent threats, aiding monitoring and patrolling efforts.
FieldKit and the Arribada Initiative aim to make the technology more accessible by developing low-cost, open-source sensor systems, while Smart Parks and Sensing Clues focus on using networked sensors to optimize protected area monitoring and management.
Most national parks don’t have basic internet or cellphone coverage as national telecommunications networks don’t typically extend to these protected areas. To provide low-power, long-range connectivity, Smart Parks deploys a range of sensors, including gate sensors, alarm systems, and animal, vehicle and people trackers, which run autonomously on solar power, consume little energy and are connected to a secure private network situated in the park itself.
The networked sensors track a wide range of information, and are able to detect human intrusions which can support anti-poaching efforts, or animal breakouts from the protected area into the community which could help preempt human-wildlife conflict.
The data is made available in or near real time in a web application, and can help inform operational decisions related to park management, wildlife conservation and local community protection, and could even be applied to ensure ranger and tourist safety.
Smart Parks technology has been deployed in protected areas around the world, and has helped contribute to the conservation of many endangered species, including orangutans, rhinos and elephants.
Gaming wildlife protection
Although it was not covered by the WILDLABS survey, games can also serve as a valuable tool to activate audiences with critical conservation issues, especially among a younger and more tech-savvy generation. Internet of Elephants, for example, develops a range of gaming and digital experiences based on scientific data to engage people who might not have otherwise held an interest in wildlife conservation.
Its products include Wildeverse, an augmented reality mobile game where players go on conservation missions in the jungle and learn how to keep apes safe, or Unseen Empire, which has turned one of the largest camera trap studies into a gaming experience. Players review real-life camera trap imagery to identify various wildlife species, and in the process learn more about the devastating impact of deforestation, poaching and other human developments on endangered wildlife, including the elusive clouded leopards.
Reducing inequalities in conservation tech
Besides highlighting the most promising tech innovations, the WILDLABS report has also identified some of the key challenges facing the conservation technology ecosystem, including competition for limited funding, duplication of efforts and insufficient capacity-building.
Importantly, the research revealed that financial and technical barriers might disproportionately affect women and people in developing countries.
“Many of the most critical conservation hotspots are also areas that are currently receiving the least support in terms of local tech capacity building,” shared Talia Speaker, WILDLABS Research Lead at WWF and co-author of the report.
Speaker warned about the problematic nature of “parachute science” which involves scientists and conservationists from high-income countries providing temporary support in developing nations and leaving after the project is finished, with no investment in local capacity-building. Without empowering local communities to use and develop conservation technologies themselves, the effectiveness and long-term sustainability of these solutions are put at risk.
To address these challenges, “the findings of this research are already feeding into a variety of WILDLABS programs,” added Speaker. “These range from fellowships that bridge the technology and conservation sectors to targeted community and capacity-building in regions like East Africa and Southeast Asia with high potential for conservation tech impact but historically limited resources for engagement with the field.”
,
Read More »Five Ways AI Is Saving Wildlife
February 27th, 2022
Courtesy of The Guardian, a look at how AI is being used in conservation:
There’s a strand of thinking, from sci-fi films to Stephen Hawking, that suggests artificial intelligence (AI) could spell doom for humans. But conservationists are increasingly turning to AI as an innovative tech solution to tackle the biodiversity crisis and mitigate climate change.
A recent report by Wildlabs.net found that AI was one of the top three emerging technologies in conservation. From camera trap and satellite images to audio recordings, the report notes: “AI can learn how to identify which photos out of thousands contain rare species; or pinpoint an animal call out of hours of field recordings – hugely reducing the manual labour required to collect vital conservation data.”
AI is helping to protect species as diverse as humpback whales, koalas and snow leopards, supporting the work of scientists, researchers and rangers in vital tasks, from anti-poaching patrols to monitoring species. With machine learning (ML) computer systems that use algorithms and models to learn, understand and adapt, AI is often able to do the job of hundreds of people, getting faster, cheaper and more effective results.
Here are five AI projects contributing to our understanding of biodiversity and species:
1. Stopping poachers
Zambia’s Kafue national park is home to more than 6,600 African savanna elephants and covers 22,400 sq km, so stopping poaching is a big logistical challenge. Illegal fishing in Lake Itezhi-Tezhi on the park’s border is also a problem, and poachers masquerade as fishers to enter and exit the park undetected, often under the cover of darkness.Stopping poachers at Kafue national park.
Automated alerts mean that just a handful of rangers are needed to provide around-the-clock surveillance. Photograph: Game Rangers International
The Connected Conservation Initiative, from Game Rangers International (GRI), Zambia’s Department of National Parks and Wildlife and other partners, is using AI to enhance conventional anti-poaching efforts, creating a 19km-long virtual fence across Lake Itezhi-Tezhi. Forward-looking infrared (FLIR) thermal cameras record every boat crossing in and out of the park, day and night.Installed in 2019, the cameras were monitored manually by rangers, who could then respond to signs of illegal activity. FLIR AI has now been trained to automatically detect boats entering the park, increasing effectiveness and reducing the need for constant manual surveillance. Waves and flying birds can also trigger alerts, so the AI is being taught to eliminate these false readings.
“There have long been insufficient resources to secure protected areas, and having people watch multiple cameras 24/7 doesn’t scale,” says Ian Hoad, special technical adviser at GRI. “AI can be a gamechanger, as it can monitor for illegal boat crossings and alert ranger teams immediately. The technology has enabled a handful of rangers to provide around-the-clock surveillance of a massive illegal entry point across Lake Itezhi-Tezhi.”
2. Tracking water loss
Brazil has lost more than 15% of its surface water in the past 30 years, a crisis that has only come to light with the help of AI. The country’s rivers, lakes and wetlands have been facing increasing pressure from a growing population, economic development, deforestation, and the worsening effects of the climate crisis. But no one knew the scale of the problem until last August, when, using ML, the MapBiomas water project released its results after processing more than 150,000 images generated by Nasa’s Landsat 5, 7 and 8 satellites from 1985 to 2020 across the 8.5m sq km of Brazilian territory. Without AI, researchers could not have analysed water changes across the country at the scale and level of detail needed. AI can also distinguish between natural and human-created water bodies.The Negro River, a major tributary of the Amazon and one of the world’s 10 largest rivers by volume, has lost 22% of its surface water. The Brazilian portion of the Pantanal, the world’s largest tropical wetland, has lost 74% of its surface water. Such losses are devastating for wildlife (4,000 species of plants and animals live in the Pantanal, including jaguars, tapirs and anacondas), people and nature.
“AI technology provided us with a shockingly clear picture,” says Cássio Bernardino, WWF-Brasil’s MapBiomas water project lead. “Without AI and ML technology, we would never have known how serious the situation was, let alone had the data to convince people. Now we can take steps to tackle the challenges this loss of surface water poses to Brazil’s incredible biodiversity and communities.”
3. Finding whales
Knowing where whales are is the first step in putting measures such as marine protected areas in place to protect them. Locating humpbacks visually across vast oceans is difficult, but their distinctive singing can travel hundreds of miles underwater. At National Oceanic and Atmospheric Association (Noaa) fisheries in the Pacific islands, acoustic recorders are used to monitor marine mammal populations at remote and hard-to-access islands, says Ann Allen, Noaa research oceanographer. “In 14 years, we’ve accumulated around 190,000 hours of acoustic recordings. It would take an exorbitant amount of time for an individual to manually identify whale vocalisations.”Google AI that recognises Humpback Whale Song bars.
AI is helping researchers in the Pacific islands recognise whale song from acoustic recordings. Photograph: Noaa
In 2018, Noaa partnered with Google AI for Social Good’s bioacoustics team to create an ML model that could recognise humpback whale song. “We were very successful in identifying humpback song through our entire dataset, establishing patterns of their presence in the Hawaiian islands and Mariana islands,” says Allen. “We also found a new occurrence of humpback song at Kingman reef, a site that’s never before had documented humpback presence. This comprehensive analysis of our data wouldn’t have been possible without AI.”4. Protecting koalas
Australia’s koala populations are in serious decline due to habitat destruction, domestic dog attacks, road accidents and bushfires. Without knowledge of their numbers and whereabouts, saving them is challenging. Grant Hamilton, associate professor of ecology at Queensland University of Technology (QUT), has created a conservation AI hub with federal and Landcare Australia funding to count koalas and other endangered animals. Using drones and infrared imaging, an AI algorithm rapidly analyses infrared footage and determines whether a heat signature is a koala or another animal. Hamilton used the system after Australia’s devastating bushfires in 2019 and 2020 to identify surviving koala populations, particularly on Kangaroo Island.“This is a gamechanger project to protect koalas,” says Hamilton. “Powerful AI algorithms are able to analyse countless hours of video footage and identify koalas from many other animals in the thick bushland. This system will allow Landcare groups, conservation groups and organisations working on protecting and monitoring species to survey large areas anywhere in Australia and send the data back to us at QUT to process it.
“We will increasingly see AI used in conservation,” he adds. “In this current project, we simply couldn’t do this as rapidly or as accurately without AI.”
5. Counting species
Saving species on the brink of extinction in the Congo basin, the world’s second-largest rainforest, is a huge task. In 2020, data science company Appsilon teamed up with the University of Stirling in Scotland and Gabon’s national parks agency (ANPN) to develop the Mbaza AI image classification algorithm for large-scale biodiversity monitoring in Gabon’s Lopé and Waka national parks.Conservationists had been using automated cameras to capture species, including African forest elephants, gorillas, chimpanzees and pangolins, which then had to be manually identified. Millions of pictures could take months or years to classify, and in a country that is losing about 150 elephants each month to poachers, time matters.
The Mbaza AI algorithm was used in 2020 to analyse more than 50,000 images collected from 200 camera traps spread across 7,000 sq km of forest. Mbaza AI classifies up to 3,000 images an hour and is up to 96% accurate. Conservationists can monitor and track animals and quickly spot anomalies or warning signs, enabling them to act swiftly when needed. The algorithm also works offline on an ordinary laptop, which is helpful in locations with no or poor internet connectivity.
“Many central African forest mammals are threatened by unsustainable trade, land-use changes and the global climate crisis,” says Dr Robin Whytock, post-doctoral research fellow at the University of Stirling. “Appsilon’s work on the Mbaza AI app enables conservationists to rapidly identify and respond to threats to biodiversity. The project started with 200 camera traps in Lopé and Waka national parks in Gabon but, since then, hundreds more have been deployed by different organisations across west and central Africa. In Gabon, the government and national parks agency are aiming to deploy cameras across the entire country. Mbaza AI can help all these projects speed up data analysis.”
,
Read More »