AI — combined with digital cameras, mobile devices, and cloud platforms — is radically transforming wildlife conservation and research efforts by optimizing data collection, monitoring, and analysis efforts. AI-powered platforms also facilitate greater collaboration among a broad range of members of the global conservation community.

Key Developments, Applications & Users

AI is used with various sensor-equipped devices, including mobile phones, trail cameras, aircraft (crewed and autonomous drones), advanced thermal-imaging systems, and audio recorders. Users range from scientists, researchers, and law enforcement to ordinary citizens interested in supporting conservation efforts. The latter is a particularly important development; such “citizen scientists” contribute large numbers of photos, videos, and audio recordings of wildlife to research and conservation efforts worldwide.

Key application domains and use cases for applying AI in wildlife conservation and research include:

• Automating species identification

• Monitoring and tracking animals

• Protecting wildlife

• Data management and analysis

Automating Species Identification

AI-powered recognition software can automatically identify species based on photographs and audio recordings, significantly speeding up the process of cataloging wildlife.

Mobile Identification Apps

The most visible example of AI-automated species identification is found in the many apps readily available for download on Apple iStore and Google Play. Apps like Merlin Bird ID (Cornell Lab), iNaturalist (iNaturalist), and Picture Nature: Animal ID (Next Vision Limited) let users identify birds and other animals by simply taking their picture or capturing an audio recording of animal vocalizations (e.g., bird tweets). The apps can also be used to connect and collaborate with communities of like-minded animal lovers, naturalists, and scientists.

Monitoring & Tracking Animals

AI enhances the ability of field cameras, drones, and crewed aircraft to monitor and track wildlife more effectively.

The Happywhale Project

Happywhale is an innovative application that combines citizen scientists with AI technology to track and study whales. Anyone, from whale-watching tourists to fishermen and scientists, can participate by submitting photos of whales they encounter to the Happywhale website. Tracking individual whales enables researchers to monitor whale population health and migration patterns and assess the impacts of environmental changes on whale populations in general.

Happywhale uses image recognition to identify whales similar to the way face recognition can identify people. It identifies individual whales based on the unique markings appearing on their tails in the form of scars, pigmentation, and other patterns recognizable by the image-recognition algorithms, including the distinctive trailing edge of the tail.

Upon identification, a whale is assigned a number, and the whale’s location, time of sighting, behavior (e.g., breaching, playing), and other relevant information are entered into a database. Happywhale has collected over 1 million photos and identified around 30,000 individual whales. This data has been used in various studies, including one that revealed a 20% decline (from 2012–2021) in the North Pacific humpback whale population due to a marine heatwave.

Identifying & Monitoring Great White Sharks

SharkEye, a project by the Benioff Ocean Initiative at the University of California Santa Barbara, uses drones to monitor and identify great white sharks in the Pacific Ocean. Flying over the ocean, drones (operated by Federal Aviation Administration [FAA]-certified pilots) shoot real-time video of the water below. This footage is analyzed by an AI application developed by Salesforce AI Research and data scientists at San Diego State University. It uses computer vision to identify sharks based on their size and shape. Upon detecting a shark, an alert is sent to local lifeguards, surf shop owners, and beachgoers who have signed up for the service. The collected data is also used to study shark behavior, monitor their movement patterns, and support other shark conservation research.

Protecting Wildlife

Protecting wildlife includes ensuring the safety of protected species from criminal elements and potential hazards (both natural and man-made.)

Poaching Prevention

AI systems, integrated with motion-detection cameras (including thermal imaging) can detect and alert game wardens to the presence of poachers entering game preserves and other protected areas. Such applications can differentiate humans from animals via the use of high-resolution images captured in real time, allowing wardens to interdict illegal hunting. Images and video can also provide evidence for prosecuting offenders.

Reducing Wildlife Deaths Caused by Fences

Old fences, particularly those made of barbed wire, are dangerous for migratory wildlife like deer and elk. Conservationists want to remove old fences to limit this danger, but this first requires locating them — a task that is difficult in the western US (and other areas of the world) where a large number of fences have been built.

Science magazine recently reported on efforts by researcher Wenjing Xu from the Senckenberg Biodiversity and Climate Research Centre and Zhongqi Miao, a scientist at Microsoft’s AI for Good Lab, to develop a deep learning model trained to identify large tracts of fences remotely using images from planes flying over southwestern Wyoming (USA). The researchers indicate that their application was approximately 70% effective in identifying fences in images. They plan to further refine their model, including to support satellite images.

Data Management & Analysis

AI applications residing on cloud platforms help optimize the huge volumes of photos, audio recordings, and other data collected from varying sources for wildlife conservation and research. They also support collaboration among researchers and scientific groups and foster the participation of citizen scientists, animal lovers, and other wildlife enthusiasts in conservation efforts.

Wildlife Insights

Wildlife Insights is a cloud platform developed by Google and various partners to streamline wildlife conservation efforts by applying advanced technology to manage, analyze, and share data captured from motion-activated cameras deployed in the wild. It was developed in collaboration with the Wildlife Conservation Society, Smithsonian Conservation Biology Institute, North Carolina Museum of Natural Sciences, WWF, Zoological Society of London, and Conservation International.

The platform uses AI to filter out non-relevant images and identify species in photos, making it easier for researchers to analyze data and make more informed decisions. Motion-triggered cameras enable researchers to capture images of wildlife in their natural habitats. These images are then uploaded to the Google Cloud, where AI models automatically identify and classify the species in the photos. This automation enables researchers to quickly generate biodiversity insights and monitor wildlife populations more efficiently.

The platform’s AI, data management, and processing power are massive. Its Cloud AI Platform Predictions tool and custom AI models reportedly can analyze up to 3.6 million photos an hour. The models are trained to distinguish between empty images and those containing wildlife and to predict the species present in the images with high accuracy. This lets researchers focus on interpreting the data and making informed decisions (as opposed to spending time preprocessing the data). With additional tools, users can create maps and charts for visualizing wildlife trends and share their findings with the global conservation community.

Conclusion

The applications covered in this Advisor by no means represent all those in use or under development. However, I believe they offer a good overview of how AI is transforming wildlife conservation by enabling more efficient and accurate monitoring, protection, and research of various species. An upcoming Advisor will examine the development of AI models for wildlife conservation applications. In the meantime, I’d like to get your opinion on using AI for wildlife conservation and research. As always, your comments will be held in strict confidence. You can email me at experts@cutter.com or call +1 510 356 7299 with your comments.