View source for AI and Wildlife Monitoring

{{#seo:

|title=AI and Wildlife Monitoring: How Artificial Intelligence Is Transforming Conservation

|description=Artificial intelligence is transforming wildlife monitoring through camera traps, bioacoustics, drones, satellite imagery, individual animal recognition, real-time alerts, and automated biodiversity surveys.

|keywords=AI wildlife monitoring, artificial intelligence conservation, wildlife conservation technology, camera traps, bioacoustics, wildlife drones, satellite wildlife monitoring, machine learning conservation, SpeciesNet, BirdNET, Wildlife Insights, animal recognition, biodiversity monitoring, anti-poaching technology

|image=File:Placeholder.png

|image_width=300

|image_height=200

|type=article}}

[[Category:Artificial Intelligence]]
[[Category:Wildlife Conservation]]
[[Category:Biodiversity Monitoring]]
[[Category:Conservation Technology]]
[[Category:Machine Learning]]
[[Category:Wildlife]]
[[Category:Environmental Science]]

**NOTOC**

== AI and Wildlife Monitoring ==

Artificial intelligence is rapidly changing how scientists, conservation organizations, governments, and wildlife managers observe animals and ecosystems. Traditionally, wildlife monitoring required researchers to spend enormous amounts of time examining photographs, listening to recordings, conducting field surveys, counting animals, or manually reviewing aerial imagery. AI can increasingly automate portions of these tasks, allowing researchers to process quantities of ecological data that would otherwise be difficult or impossible to examine manually.

Modern wildlife-monitoring systems combine artificial intelligence with camera traps, microphones, drones, thermal cameras, satellites, underwater video, sonar, animal-borne sensors, and other technologies. Machine-learning systems can identify species, count animals, recognize individuals, classify behavior, detect animal sounds, estimate population density, locate endangered species, and sometimes transmit alerts almost immediately when wildlife or potential threats are detected.

The result is not simply faster data processing. AI is gradually changing wildlife monitoring from a system in which researchers periodically collect and analyze observations into one capable of providing increasingly continuous information about animals and ecosystems.

=== Camera Traps and Computer Vision ===

Camera traps have become one of the most important sources of data for artificial-intelligence wildlife monitoring. Motion-activated cameras can operate for weeks or months in forests, grasslands, deserts, wetlands, and other environments, photographing animals that might rarely be observed directly by researchers.

Their success creates another problem: an individual conservation project can accumulate hundreds of thousands or even millions of photographs.

Computer-vision systems can automatically examine these images and determine whether they contain animals, people, vehicles, or nothing of interest. More advanced models can identify species, count individuals, estimate their distance from cameras, recognize behavior, and sometimes distinguish individual animals.

Systems such as MegaDetector, SpeciesNet, Wildlife Insights, Conservation AI, and numerous specialized research models demonstrate the growing ability of machine learning to process large wildlife-image collections.

Recent research has also focused on making these systems more practical for conservationists. Lightweight models can run directly on low-powered field computers, while specialized workflows allow ecologists without extensive computer-programming expertise to train models for their own wildlife populations.

AI can also help solve one of the persistent difficulties of camera-trap research: recognizing species at new camera locations. Changes in vegetation, lighting, weather, camera angle, and background scenery can cause models trained in one location to perform poorly elsewhere. Researchers are therefore developing models that rely less on background information and more on the physical characteristics of the animals themselves.

=== Identifying Individual Animals ===

AI wildlife monitoring is increasingly moving beyond identifying species toward recognizing individual animals.

Many animals possess distinctive physical patterns. Leopards and jaguars have unique spots, zebras have unique stripes, giraffes have characteristic coat patterns, and some whales can be distinguished by markings, scars, or other physical characteristics.

Computer-vision systems can compare these patterns across thousands of photographs and suggest matches between images.

Individual recognition is valuable because conservation scientists frequently need to know more than whether a species exists in an area. Recognizing individuals can help estimate population size, track movement, measure survival, determine home ranges, examine dispersal, and study relationships between animals.

Research has applied AI-based individual recognition to animals including giraffes, leopards, tigers, bears, snow leopards, primates, birds, fish, skates, and other species.

These techniques could eventually reduce reliance on physically capturing, tagging, or marking animals, although conventional tracking and identification methods remain important.

=== Bioacoustics and Automated Wildlife Listening ===

Wildlife does not have to be visible to be monitored.

Birds, frogs, bats, whales, dolphins, insects, elephants, primates, and numerous other animals produce distinctive sounds. Networks of autonomous recording devices can collect thousands of hours of audio from forests, oceans, wetlands, grasslands, and other ecosystems.

Artificial intelligence makes it possible to search these enormous sound collections automatically.

BirdNET is one prominent example. Deep-learning systems can identify bird vocalizations from recordings and have increasingly been adapted to other wildlife-monitoring applications.

Researchers are developing AI systems for detecting frogs, bats, whales, marine mammals, insects, endangered birds, invasive species, and other animals from sound.

Bioacoustic AI can help scientists determine where animals occur, when they are active, how their distribution changes, and whether ecological communities are recovering or declining.

Acoustic monitoring is particularly valuable for species that are difficult to see. Some animals may remain hidden in dense vegetation while continuing to vocalize, making sound a more reliable indicator of their presence than visual observations.

Automated listening also allows scientists to study entire soundscapes rather than individual species. Changes in the diversity or frequency of biological sounds can provide evidence of broader ecological change.

=== Drones and Aerial Wildlife Monitoring ===

Drones provide another rapidly developing platform for AI wildlife monitoring.

Traditional aerial wildlife surveys frequently require researchers to fly over large areas in aircraft while visually counting animals. These surveys can be expensive, dangerous, and difficult to repeat frequently.

Drones can collect high-resolution photographs, video, and thermal imagery at lower altitudes. Artificial-intelligence systems can then search the resulting imagery for wildlife.

AI-assisted drone research has been applied to mammals, birds, marine animals, endangered species, and animals involved in human-wildlife conflicts.

Deep-learning systems can distinguish wildlife from complex backgrounds and can sometimes detect small animals that are difficult for human observers to locate.

Thermal cameras expand these possibilities by detecting body heat rather than relying entirely on visible-light imagery. Combining thermal and ordinary RGB imagery can improve detection under difficult environmental conditions.

Researchers are also developing autonomous wildlife-monitoring drones capable of navigating landscapes, locating animals, tracking them, and adjusting survey routes with limited human intervention.

=== Satellites and Wildlife Monitoring from Space ===

Artificial intelligence is also expanding wildlife monitoring far beyond what can be observed from the ground.

High-resolution satellites increasingly produce imagery detailed enough to identify certain large animals or animal groups. AI can search these enormous geographic datasets far more rapidly than human analysts.

Satellite monitoring is particularly attractive for remote environments where conventional surveys are difficult, expensive, or dangerous.

Projects are investigating satellite and geospatial AI for detecting whales, including critically endangered North Atlantic right whales and Cook Inlet beluga whales.

Satellite imagery can also provide information about habitat rather than animals themselves. Machine-learning systems can analyze vegetation, forests, wetlands, coastlines, water conditions, and other environmental characteristics that influence wildlife populations.

Combining animal observations with habitat information could eventually allow conservationists to monitor both wildlife populations and the ecological conditions supporting them.

=== Marine and Underwater Monitoring ===

AI wildlife monitoring is increasingly important below the water as well as above it.

Underwater cameras, remotely operated vehicles, sonar systems, hydrophones, and satellite observations generate large quantities of information about fish, marine mammals, seabirds, and aquatic ecosystems.

Computer-vision systems can identify and count fish from underwater video. Other systems estimate body size, track animals through video, classify marine vegetation, or distinguish species that appear visually similar.

Sonar-based machine learning can detect animals in water where ordinary cameras perform poorly.

Acoustic AI is particularly important for whales and dolphins because these animals communicate over long distances and may be heard even when they cannot be seen.

Together, these technologies are creating increasingly automated systems for observing marine ecosystems that were historically among the most difficult environments for scientists to monitor continuously.

=== Monitoring Animal Behavior ===

Artificial intelligence is also expanding the study of animal behavior.

Traditional behavioral research often requires scientists to watch hours of video and manually record activities such as feeding, resting, walking, swimming, grooming, fighting, or interacting with other animals.

Computer vision can increasingly automate portions of this process.

AI systems can follow animals through video, recognize body positions, identify behaviors, and reconstruct movement patterns.

Researchers are experimenting with systems capable of identifying complex behaviors among endangered wildlife, monitoring swimming birds, examining primate facial movements, and measuring animal activity over long periods.

These applications could allow scientists to study behavior continuously rather than relying on relatively short periods of direct human observation.

=== Human-Wildlife Conflict ===

One of the most immediate applications of AI wildlife monitoring involves protecting both wildlife and people.

Large animals such as elephants, tigers, leopards, and other wildlife can sometimes approach farms, roads, livestock areas, or communities. Traditional monitoring may not provide warnings quickly enough to prevent dangerous encounters.

AI-equipped cameras can automatically recognize animals and transmit alerts when they approach populated areas.

Systems developed for elephants, for example, can combine cameras, computer vision, GPS information, and communication networks to warn conservation teams or communities when animals are approaching.

Real-time tiger-monitoring systems have similarly demonstrated how AI cameras can detect endangered predators and rapidly transmit information.

Such systems illustrate an important transition in conservation technology: monitoring can move from documenting events after they happen toward generating information quickly enough to influence what happens next.

=== Anti-Poaching and Wildlife Protection ===

The same principle can be applied to wildlife crime.

Camera systems equipped with artificial intelligence can recognize not only animals but also people and vehicles entering protected landscapes.

Thermal cameras can operate at night, when many conventional surveillance systems are less effective.

In Kenya, AI-assisted thermal surveillance has been used in rhino conservation, allowing monitoring systems to identify people, vehicles, and wildlife and alert ranger teams.

Edge computing can make these systems more useful in remote areas. Instead of storing photographs until researchers retrieve a memory card weeks or months later, the camera itself can analyze images and transmit significant detections.

This turns camera traps into potential real-time conservation sentinels.

=== Edge AI and Remote Monitoring ===

One important technological development is the movement of artificial intelligence from distant data centers directly into wildlife-monitoring devices.

Traditional camera traps simply collect information. Researchers later retrieve memory cards and process the photographs.

An edge-AI camera can analyze those photographs immediately.

If nothing important is detected, the image may not need to be transmitted. If an elephant, tiger, invasive species, poacher, or other important target appears, the system can prioritize the information and send an alert.

This is particularly valuable in remote conservation areas where internet connectivity is limited.

Solar-powered systems, lightweight neural networks, satellite communications, and specialized low-energy computers are increasingly being combined to create wildlife-monitoring stations capable of operating autonomously for long periods.

Projects such as Sentinel and SPARROW illustrate this movement toward distributed conservation intelligence.

=== AI, Citizen Science and Conservation Platforms ===

Artificial intelligence is also becoming integrated into broader networks for collecting and sharing biodiversity information.

Wildlife Insights demonstrates how standardized camera-trap information can be uploaded, processed, analyzed, and shared through a common platform.

Citizen-science observations provide another enormous source of biodiversity information. AI can help classify images and recordings submitted by volunteers while researchers verify difficult or unusual observations.

Conservation platforms are also increasingly connecting different kinds of sensors.

Rather than treating camera traps, acoustic recorders, tracking devices, drones, and satellite imagery as completely separate sources of information, emerging systems attempt to combine them into integrated ecological-monitoring networks.

The long-term goal is a more complete picture of ecosystems assembled from many different forms of evidence.

=== Human-in-the-Loop Artificial Intelligence ===

Despite rapid progress, wildlife AI does not eliminate the need for human expertise.

Artificial-intelligence models make mistakes. Similar-looking species may be confused, animals can be partially hidden, unusual lighting can distort images, background noise can interfere with acoustic recognition, and models trained in one ecosystem may perform poorly in another.

Rare species create another fundamental problem. Machine-learning systems typically learn best when thousands of examples are available, while endangered species may appear in only a handful of photographs or recordings.

Researchers are responding with few-shot learning, transfer learning, active learning, data augmentation, vision-language models, and human-in-the-loop systems.

Active learning allows an AI system to identify the examples about which it is most uncertain and ask humans to review those cases. This concentrates expert effort where it is most useful.

Human-in-the-loop conservation therefore represents an important middle ground. AI performs repetitive large-scale screening while scientists, local experts, trackers, volunteers, and conservation managers examine ambiguous or important observations.

=== Challenges and Limitations ===

The growing role of artificial intelligence in wildlife monitoring also creates important scientific and practical challenges.

AI models must be validated carefully before their results are used to estimate wildlife populations or guide conservation policy. High classification accuracy alone does not guarantee that ecological conclusions will be correct.

Models may perform differently when moved to new geographic locations or when weather, vegetation, camera equipment, or animal appearance changes.

Training datasets may contain thousands of photographs of common species but very few examples of rare species.

Remote systems also face practical limitations involving battery life, communications, equipment cost, weather resistance, storage, and computing capacity.

False detections can overwhelm monitoring teams, while missed detections can be especially serious when systems are being used for endangered species, poaching alerts, or human-wildlife conflict.

Another challenge is ensuring that increasingly sophisticated technology remains accessible to conservation organizations with limited technical or financial resources.

For these reasons, AI should generally be understood as a tool that expands human monitoring capabilities rather than a replacement for ecologists, wildlife managers, local communities, trackers, or field researchers.

=== The Future of AI Wildlife Monitoring ===

Wildlife monitoring appears to be moving toward increasingly connected systems.

Camera traps can provide images. Microphones can provide sound. Drones can provide aerial observations. Satellites can monitor enormous geographic areas. Animal-borne sensors can provide movement information. Environmental sensors can record weather and habitat conditions.

Artificial intelligence provides a means of combining and interpreting this growing stream of information.

Future conservation systems may continuously detect animals, recognize individuals, interpret behavior, map movements, identify ecological changes, and alert conservation teams when unusual events occur.

Foundation models and large pretrained wildlife models could also reduce the need to develop a completely new AI system for every species or conservation project.

At the same time, human expertise will remain essential for checking results, understanding ecological context, deciding what actions should be taken, and determining which conservation goals technology should serve.

=== Conclusion ===

Artificial intelligence is transforming wildlife monitoring by allowing conservationists to analyze ecological information at scales that were previously impractical.

Camera traps equipped with computer vision can identify and count animals. Acoustic networks can recognize wildlife from sound. Drones and satellites can survey remote landscapes. Underwater systems can monitor fish and marine mammals. Individual-recognition algorithms can follow particular animals over time. Edge-AI systems can transform passive sensors into real-time warning networks.

These technologies have applications ranging from fundamental ecological research to population surveys, endangered-species conservation, human-wildlife conflict prevention, habitat monitoring, and anti-poaching operations.

The greatest value of AI may not be that it replaces traditional wildlife research, but that it dramatically increases how much information researchers can examine and how quickly that information can become useful.

As cameras, acoustic sensors, satellites, drones, communications networks, and artificial intelligence become increasingly integrated, wildlife monitoring is moving toward a future in which ecosystems can be observed more continuously and at much larger scales.

The central challenge will be ensuring that greater technological capability produces better conservation decisions. Artificial intelligence can reveal patterns hidden within millions of photographs, recordings, and observations, but protecting wildlife ultimately depends on how people choose to use that knowledge.

**TOC**




=== General AI and Foundational Wildlife Monitoring ===

[https://www.theguardian.com/environment/2026/aug/27/how-decoding-beluga-whales-chitchat-may-save-them-and-teach-us-more-about-ourselves | Phoebe Weston | The Guardian | 2026-08-27]
 ''How decoding beluga whales' chitchat may save them'' examines AI-assisted analysis of whale communication and how understanding vocal cultures could improve marine conservation.

[https://www.axios.com/local/portland/2026/08/26/spotted-owl-functional-extinction-study | Axios Portland | Axios | 2026-08-26]
 Reporting on northern spotted owl declines describes researchers using automated listening devices and machine learning to assess vocal activity and population persistence across the Pacific Northwest.

[https://www.frontiersin.org/journals/conservation-science/articles/10.3389/fcosc.2026.1837914/full | Delia Velasco-Montero et al. | Frontiers in Conservation Science | 2026-08-24]
 ''Bridging the edge–cloud gap: adaptive AI for robust image and audio wildlife monitoring'' reviews AI systems that analyze both camera-trap imagery and wildlife sounds, with emphasis on edge computing, continual learning, generalization, and hybrid cloud systems.

[https://www.sciencedirect.com/science/article/pii/S2590123026036248 | Multiple authors | Results in Engineering | 2026-08-24]
 ''Deep Learning for Environmental Monitoring and Conservation'' systematically reviews more than 100 studies using deep learning for conservation, including wildlife detection, remote sensing, ecological monitoring, CNNs, Transformers, and YOLO models.

[https://www.sciencedirect.com/science/article/pii/S0048969726005905 | Prakash Palanivelu Rajmohan et al. | Science of the Total Environment | 2026-08-01]
 ''AI and computer vision for wildlife identification in camera trap images'' finds that locally fine-tuning the global SpeciesNet model can exceed 95 percent F1 scores with only several hundred training images per species.

[https://www.sciencedirect.com/science/article/pii/S2351989426001721 | Nuo Xu et al. | Global Ecology and Conservation | 2026-08]
 ''Automatic re-identification of terrestrial mammals using deep learning and camera trap images'' presents ARNet, which combines vision-language information and multi-level image features to distinguish individual mammals.

[https://www.sciencedirect.com/science/article/pii/S157495412600333X | David Bolduc et al. | Ecological Informatics | 2026-08]
 ''Speeding up image annotation and AI-model training for wildlife images'' combines Timelapse, MegaDetector, SAM2, YOLO, and R to reduce manual annotation effort and make custom wildlife models easier for ecologists to build.

[https://link.springer.com/article/10.1007/s42977-026-00322-5 | Sándor Zsebők | Biologia Futura | 2026-06-09]
 ''Listening forward: emerging roles of bioacoustics in ecology, evolution, and conservation'' examines autonomous recorders, transfer learning, unsupervised learning, edge computing, and explainable AI for monitoring wildlife through sound.

[https://www.sciencedirect.com/science/article/pii/S1574954126002529 | Ahmed Shahabaz et al. | Ecological Informatics | 2026-06]
 ''From relative to metric'' applies AI monocular-depth models to camera-trap imagery so animal distances can be estimated automatically for population-density surveys.

[https://www.sciencedirect.com/science/article/pii/S1574954126001718 | Multiple authors | Ecological Informatics | 2026-06]
 ''Foundation models for bioacoustics – A comparative review'' evaluates large pretrained AI audio models for animal-sound recognition and biodiversity monitoring.

[https://www.nature.com/articles/s41598-026-54115-w | Shahrzad Gholami et al. | Scientific Reports | 2026-05-30]
 ''An accurate, efficient, and accessible AI-powered solution for wildlife re-identification in conservation'' introduces GIRAFFE, an open-source human-in-the-loop system capable of matching individual giraffes and other patterned wildlife.

[https://www.nature.com/articles/s41598-026-48229-4 | Tinao Petso et al. | Scientific Reports | 2026-04-20]
 ''Improving wildlife track classification through human-in-the-loop method and explainable AI'' combines expert animal trackers with machine learning and visual explanation tools to identify species from footprints.

[https://www.nature.com/collections/fgcaibhggj | Scientific Reports Editors | Scientific Reports | 2026]
 ''Automated biodiversity monitoring'' describes the rapidly developing integration of camera traps, acoustic sensors, remote sensing, AI, and edge computing for continuous ecosystem monitoring.

[https://ijirp.org/index.php/files/article/view/137 | Salman Arafath Mohammed | International Journal of Integrated Research and Practice | 2026]
 ''Artificial Intelligence in Wildlife Tracking and Conservation'' surveys computer vision, bioacoustics, predictive modeling, drone surveillance, telemetry, and habitat monitoring applications.

[https://birdnet.cornell.edu/ | Cornell Lab of Ornithology and Chemnitz University of Technology | BirdNET | 2026]
 BirdNET's research platform illustrates the mature use of AI for wildlife monitoring, with deep-learning models capable of recognizing thousands of species from sound at large geographic scales.

[https://www.sciencedirect.com/science/article/pii/S3050472425000103 | Multiple authors | Water & Ecology | 2025-08]
 ''Artificial Intelligence in Aquatic Biology'' examines machine-learning methods for identifying aquatic species, evaluating habitat conditions, and detecting conservation threats.

[https://www.mdpi.com/2673-7159/4/4/41 | Multiple authors | Conservation | 2024-11-11]
 ''Harnessing Artificial Intelligence for Wildlife Conservation'' reviews Conservation AI and related computer-vision tools that detect animals, people, and potential poaching activity from visible and thermal imagery.

[https://link.springer.com/article/10.1007/s42977-023-00200-4 | Zoltán Barta | Biologia Futura | 2024-01-16]
 ''Deep learning in terrestrial conservation biology'' reviews AI applied to camera traps, acoustic recordings, satellite imagery, individual identification, and other biodiversity data.

[https://www.nature.com/articles/s41467-022-27980-y | Devis Tuia et al. | Nature Communications | 2022]
 ''Perspectives in machine learning for wildlife conservation'' explains how machine learning can unlock ecological information contained in camera traps, drones, satellites, audio recorders, and biologging sensors.

[https://academic.oup.com/bioscience/article/71/10/1038/6322306 | Multiple authors | BioScience | 2021]
 ''Comprehensive Overview of Technologies for Species and Habitat Monitoring and Conservation'' surveys conservation sensors, camera traps, drones, telemetry, biologging, algorithms, and artificial intelligence.

[https://www.worldwildlife.org/news/stories/new-technology-and-collaboration-could-transform-wildlife-monitoring/ | WWF | WWF | 2019]
 ''New technology and collaboration could transform wildlife monitoring'' describes how AI-assisted camera-trap processing can accelerate work on highly threatened species such as the Javan rhino.

[https://blog.google/innovation-and-ai/infrastructure-and-cloud/google-cloud/cloud-automl-making-ai-accessible-every-business/ | Google Cloud | Google | 2018-01]
 Google's ''Cloud AutoML'' announcement includes work with the Zoological Society of London to automate identification of elephants, lions, giraffes, and other species in enormous camera-trap datasets.

[https://www.worldwildlife.org/our-work/wildlife/wildlife-technology/wildlife-insights/ | WWF and Wildlife Insights partners | WWF | n.d.]
 Wildlife Insights demonstrates how standardized camera-trap databases and machine learning can turn millions of raw wildlife photographs into information useful for management and conservation decisions.

[https://www.worldwildlife.org/our-work/wildlife/camera-traps/ | WWF | WWF | n.d.]
 WWF's overview of camera traps explains the monitoring technology that now supplies much of the image data used to train and operate modern wildlife AI systems.

=== Camera Traps, Computer Vision and Automated Population Monitoring ===

[https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/1365-2664.70370 | Daniel Thornton et al. | Journal of Applied Ecology | 2026-05-06]
 ''Identification of camera trap images by artificial intelligence and human experts produces similar multi-species occupancy models'' tests whether fully automated AI identifications can produce ecological occupancy estimates comparable to those derived from expert-reviewed images.

[https://www.sciencedirect.com/science/article/pii/S1574954125004662 | Anne C. Eichholtzer et al. | Ecological Informatics | 2025-12]
 ''Integrating AI technologies and citizen science to fast-track small and ectothermic animal monitoring'' combines specialized cameras, volunteers, and YOLO detection for reptiles, amphibians, arthropods, and other small animals.

[https://www.sciencedirect.com/science/article/pii/S1574954125002675 | Multiple authors | Ecological Informatics | 2025-12]
 ''Body-part-based individual feral cat identification from camera trap images using deep learning'' investigates whether flanks and other body regions can distinguish individual invasive feral cats.

[https://www.sciencedirect.com/science/article/pii/S1574954125005291 | Multiple authors | Ecological Informatics | 2025-12]
 A deep-learning pipeline combines monocular depth estimation, automatic animal detection, and camera calibration to estimate both distance and body height from camera-trap photographs.

[https://www.sciencedirect.com/science/article/pii/S1574954125004029 | Multiple authors | Ecological Informatics | 2025-12]
 Lightweight AI models classify African ungulates from their footprints, creating a non-invasive monitoring method potentially useful for biodiversity surveys and anti-poaching.

[https://www.sciencedirect.com/science/article/pii/S1574954125001293 | Multiple authors | Ecological Informatics | 2025-11]
 ''A photogrammetric approach to the estimation of distance to animals in camera trap images'' automates animal-distance measurements needed for abundance and density estimation.

[https://www.sciencedirect.com/science/article/pii/S1574954125001578 | Multiple authors | Ecological Informatics | 2025-11]
 ''Wild ActionFormer'' uses self-supervised video learning to automatically recognize behaviors performed by 11 endangered wildlife species recorded in China's Wolong Nature Reserve.

[https://www.sciencedirect.com/science/article/pii/S1574954125001888 | Multiple authors | Ecological Informatics | 2025-11]
 A desert bighorn sheep study shows how locally tailored training datasets can make AI wildlife classifiers more reliable when deployed at previously unseen monitoring sites.

[https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.70144 | Multiple authors | Methods in Ecology and Evolution | 2025-08-25]
 ''An active ensemble classifier for detecting animal sequences from global camera trap data'' uses movement information and multiple classifiers to recognize animals despite changing habitats, weather, and illumination.

[https://doi.org/10.1155/int/1561380 | Siyabonga Mamapule et al. | International Journal of Intelligent Systems | 2025-07-25]
 ''Automatic Identification and Counting of South African Animal Species in Camera Traps Using Deep Learning'' applies CNN and YOLO models to automatically identify and count buffalo, elephants, rhinos, and zebras.

[https://www.sciencedirect.com/science/article/pii/S1574954125001001 | Multiple authors | Ecological Informatics | 2025-07]
 ''Metric learning unveiling disparities'' uses metric learning and clustering to detect false-trigger camera images, potentially reducing processing loads on resource-constrained monitoring systems.

[https://www.sciencedirect.com/science/article/pii/S0925231225004989 | Multiple authors | Neurocomputing | 2025-06-14]
 ''Enhancing generalization in camera trap image recognition'' adapts vision-language models such as CLIP to reduce dependence on background cues and improve recognition at unfamiliar camera locations.

[https://wildlife.onlinelibrary.wiley.com/doi/10.1002/wsb.1598 | Taylor L. Kaltenbach | Wildlife Society Bulletin | 2025-05-27]
 ''Can edge AI mitigate environmental effects on camera trap performance?'' field-tests an AI-enabled camera against conventional camera traps in the Greater Yellowstone Ecosystem.

[https://www.nature.com/articles/s41598-025-90249-z | Margarita Mulero-Pázmány et al. | Scientific Reports | 2025-05-09]
 ''Addressing significant challenges for animal detection in camera trap images'' uses a two-stage system of general and specialist AI models to distinguish visually similar wildlife species.

[https://doaj.org/article/1261f064f75d443aaa3c073f9a950951 | Naoya Noguchi et al. | Ecological Informatics | 2025-05]
 ''Efficient wildlife monitoring: Deep learning-based detection and counting of green turtles in coastal areas'' applies automated vision to detect and count green sea turtles from monitoring imagery.

[https://www.sciencedirect.com/science/article/pii/S157495412500055X | Multiple authors | Ecological Informatics | 2025-05]
 An ensemble-learning system automatically removes common species from camera-trap datasets while minimizing the risk that rare species will be accidentally discarded.

[https://www.sciencedirect.com/science/article/pii/S1574954124004916 | L. Wang et al. | Ecological Informatics | 2025-03]
 ''DeLoCo'' investigates how environmental and camera-location context can be used alongside animal appearance to improve automated wildlife identification.

[https://www.sciencedirect.com/science/article/pii/S1470160X25001566 | Multiple authors | Ecological Indicators | 2025-02]
 Deep learning is used to re-identify individual Amur tigers from camera traps, supporting population monitoring, home-range analysis, and studies of dispersal in Northeast Tiger and Leopard National Park.

[https://besjournals.onlinelibrary.wiley.com/doi/full/10.1002/2688-8319.70167 | Multiple authors | Ecological Solutions and Evidence | 2025]
 ''DeepForestVision'' develops a deployable deep-learning system specifically for identifying wildlife in photographs and videos from African tropical forests.

[https://www.mdpi.com/1424-8220/25/23/7331 | Multiple authors | Sensors | 2025]
 ''An Improved Lightweight Model for Protected Wildlife Detection in Camera Trap Images'' develops YOLO11-APS for accurate species detection on computing-constrained edge devices.

[https://doi.org/10.1002/ece3.70544 | Luca Petroni et al. | Ecology and Evolution | 2024-12-12]
 ''An Ecologist-Friendly R Workflow for Expediting Species-Level Classification of Camera Trap Images'' provides a practical R and YOLO workflow for researchers without extensive computer-vision programming experience.

[https://www.sciencedirect.com/science/article/pii/S1574954124003571 | Delia Velasco-Montero et al. | Ecological Informatics | 2024-11]
 ''Reliable and efficient integration of AI into camera traps for smart wildlife monitoring'' demonstrates a roughly $100 smart camera capable of running and continually improving AI models in the field.

[https://www.sciencedirect.com/science/article/pii/S1574954124003479 | Multiple authors | Ecological Informatics | 2024-11]
 ''Metadata augmented deep neural networks for wild animal classification'' shows that location, temperature, time, and other metadata can improve species classification when wildlife images are difficult to interpret.

[https://www.nature.com/articles/s41598-024-73631-1 | Bao XiaoAn et al. | Scientific Reports | 2024-10-09]
 ''Wildlife target detection based on improved YOLOX-s network'' develops a wildlife detector designed to remain effective under rain, nighttime conditions, infrared imagery, and complex backgrounds.

[https://www.sciencedirect.com/science/article/pii/S1574954124002474 | Multiple authors | Ecological Informatics | 2024-09]
 ''Adaptive image processing embedding to make the ecological tasks of deep learning more robust on camera traps images'' uses AI-controlled image enhancement to improve classification in poor lighting and difficult field conditions.

[https://www.sciencedirect.com/science/article/pii/S1574954124001201 | Multiple authors | Ecological Informatics | 2024-07]
 A semi-automated workflow removes poor-quality and empty time-lapse photographs, detects animals, crops detections, and classifies wildlife while reducing manual review.

[https://zslpublications.onlinelibrary.wiley.com/doi/abs/10.1002/rse2.412 | Gaspard Dussert et al. | Remote Sensing in Ecology and Conservation | 2024-06-16]
 ''Being confident in confidence scores'' examines whether the probability scores produced by wildlife deep-learning models can actually be trusted when camera-trap classifications feed into ecological analyses.

[https://www.sciencedirect.com/science/article/pii/S1574954124000694 | Multiple authors | Ecological Informatics | 2024-05]
 ''A systematic study on transfer learning'' explores how pretrained deep networks can automatically remove the enormous number of empty images generated by camera traps.

[https://www.sciencedirect.com/science/article/pii/S1574954124000086 | Multiple authors | Ecological Informatics | 2024-05]
 ''Benchmarking wild bird detection in complex forest scenes'' compares eight deep-learning detectors on camera-trap images of 15 bird species living in visually difficult forest environments.

[https://www.sciencedirect.com/science/article/pii/S157495412300482X | Multiple authors | Ecological Informatics | 2024-03]
 A solar-powered custom camera trap runs deep neural networks locally and uses explainable-AI techniques to show researchers why particular wildlife classifications were made.

[https://zslpublications.onlinelibrary.wiley.com/doi/abs/10.1002/rse2.367 | Multiple authors | Remote Sensing in Ecology and Conservation | 2024]
 ''Automated visitor and wildlife monitoring with camera traps and machine learning'' tests MegaDetector on more than 300,000 images containing animals, people, and vehicles from multiple regions.

[https://besjournals.onlinelibrary.wiley.com/doi/10.1111/2041-210X.14254 | Multiple authors | Methods in Ecology and Evolution | 2024]
 ''Sherlock'' provides a relatively low-resource computer-vision tool capable of filtering large camera-trap collections without requiring powerful computing equipment.

[https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/cvi2.12294 | Multiple authors | IET Computer Vision | 2024]
 Deep neural networks classify both mammals and individual bird species in European camera-trap imagery while also predicting genus, family, order, and higher taxonomic categories.

[https://www.sciencedirect.com/science/article/pii/S1877050924006148 | Multiple authors | Procedia Computer Science | 2024]
 ''Animal Identity Recognition using Object Detection Techniques'' evaluates YOLOv5 and Detectron2 for recognizing gorillas, monkeys, and other animals in difficult visual environments.

[https://www.sciencedirect.com/science/article/pii/S1574954123002601 | Multiple authors | Ecological Informatics | 2023-11]
 ''Automated wildlife image classification: An active learning tool for ecological applications'' shows how active learning can concentrate human labeling effort on the images most useful for improving a model.

[https://www.sciencedirect.com/science/article/pii/S1574954123002911 | Multiple authors | Ecological Informatics | 2023-11]
 An ensemble-learning method automatically separates wildlife photographs from images containing people and livestock, substantially reducing manual camera-trap processing.

[https://www.sciencedirect.com/science/article/pii/S1574954123002868 | Multiple authors | Ecological Informatics | 2023-11]
 ''Evaluating a tandem human-machine approach'' measures when AI suggestions genuinely accelerate human wildlife-image annotation and when incorrect predictions instead slow experts down.

[https://www.sciencedirect.com/science/article/abs/pii/S1574954123002662 | Multiple authors | Ecological Informatics | 2023-11]
 A few-shot learning method generates additional training imagery through style-transfer techniques to improve recognition of rare wildlife represented by very small datasets.

[https://www.sciencedirect.com/science/article/pii/S1574954123000894 | Multiple authors | Ecological Informatics | 2023-09]
 ''Bag of tricks for long-tail visual recognition of animal species'' tackles a core conservation-AI problem: common species have thousands of photographs while rare species may have only a few.

[https://www.sciencedirect.com/science/article/pii/S1574954123001796 | Multiple authors | Ecological Informatics | 2023-09]
 A semi-automatic R-based workflow integrates AI classification with quality control and data management for long-term small-mammal camera-trap programs.

[https://www.mdpi.com/2673-2688/4/3/31 | Haoyu Chen et al. | AI | 2023-07-31]
 ''Applying Few-Shot Learning for In-the-Wild Camera-Trap Species Classification'' examines methods that learn to recognize species when only a small number of labeled examples are available.

[https://www.sciencedirect.com/science/article/pii/S1574954123001243 | Multiple authors | Ecological Informatics | 2023]
 ''DeepWILD'' uses object detection to locate, identify, and count wildlife appearing in camera-trap videos from a French national park.

[https://doi.org/10.1111/2041-210x.14031 | Norman et al. | Methods in Ecology and Evolution | 2023]
 ''Can CNN-based species classification generalise across variation in habitat within a camera trap survey?'' tests whether wildlife classifiers remain reliable across forests experiencing different disturbance levels.

[https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.14044 | Vélez et al. | Methods in Ecology and Evolution | 2023]
 ''An evaluation of platforms for processing camera-trap data using artificial intelligence'' compares Conservation AI, MegaDetector, MLWIC2, Wildlife Insights, Camelot, and Timelapse.

[https://www.sciencedirect.com/science/article/pii/S1574954122003260 | Multiple authors | Ecological Informatics | 2022-12]
 ''Human vs. machine: Detecting wildlife in camera trap images'' tests Microsoft's MegaDetector against human reviewers and identifies conditions where each approach performs better.

[https://www.sciencedirect.com/science/article/abs/pii/S1574954122002102 | Multiple authors | Ecological Informatics | 2022-11]
 ''Class incremental learning for wildlife biodiversity monitoring'' develops AI capable of adding newly observed species without completely forgetting species learned earlier.

[https://www.sciencedirect.com/science/article/pii/S1574954122001066 | Multiple authors | Ecological Informatics | 2022-07]
 ''Motion vectors and deep neural networks for video camera traps'' describes a low-power Raspberry Pi camera system capable of filtering video and detecting animals in real time.

[https://journal.uni-mate.hu/index.php/columella/article/view/3231 | Multiple authors | Columella | 2022]
 ''A narrative review on the use of camera traps and machine learning in wildlife research'' summarizes ecological applications and major limitations of machine-learning-assisted camera trapping.

[https://doi.org/10.1111/2041-210x.13120 | Michael A. Tabak et al. | Methods in Ecology and Evolution | 2019]
 ''Machine learning to classify animal species in camera trap images'' trains convolutional neural networks on more than three million wildlife images collected across North America.

[https://www.nature.com/articles/s41598-019-44565-w | Multiple authors | Scientific Reports | 2019]
 ''Insights and approaches using deep learning to classify wildlife'' explains how convolutional neural networks recognize wildlife and reports classification of 20 African species.

[https://pmc.ncbi.nlm.nih.gov/articles/PMC6016780/ | Mohammad Sadegh Norouzzadeh et al. | Proceedings of the National Academy of Sciences | 2018]
 ''Automatically identifying, counting, and describing wild animals in camera-trap images with deep learning'' demonstrates automated analysis of the 3.2-million-image Snapshot Serengeti dataset.

[https://www.sciencedirect.com/science/article/pii/S1574954116302047 | Alexander Gómez Villa et al. | Ecological Informatics | 2017-09]
 ''Towards automatic wild animal monitoring'' is an early demonstration of very deep convolutional neural networks for automatically recognizing species in camera-trap imagery.

=== Drones, Marine Monitoring, Satellites and Remote Sensing ===

[https://www.sciencedirect.com/science/article/pii/S2590005626004698 | Multiple authors | Array | 2026-08-19]
 ''A Hierarchical Hybrid UAV Navigation Framework for Energy-Efficient and Safety-Critical Wildlife Monitoring Using Edge AI'' investigates autonomous navigation for remote wildlife-monitoring drones.

[https://www.sciencedirect.com/science/article/pii/S1574954126002189 | Multiple authors | Ecological Informatics | 2026-06]
 ''Low-disturbance UAV-AI monitoring of the endangered Przewalski's gazelle'' combines vertical-takeoff drones and YOLOv11 to count gazelles while reducing disturbance to the animals.

[https://www.mdpi.com/1424-8220/26/11/3436 | Multiple authors | Sensors | 2026-05-29]
 ''YOLIP'' combines YOLO and CLIP concepts to improve wildlife recognition from aerial imagery across scales and complicated natural backgrounds.

[https://www.sciencedirect.com/science/article/pii/S1574954126001160 | Multiple authors | Ecological Informatics | 2026-05]
 ''UAV-based deep learning for biodiversity monitoring'' reviews drones, remote sensing, Transformers, graph neural networks, generative models, and self-supervised learning for biodiversity surveys.

[https://www.sciencedirect.com/science/article/pii/S1574954126001032 | Multiple authors | Ecological Informatics | 2026-05]
 ''FDM-YOLO'' fuses thermal-infrared and RGB drone imagery to improve real-time detection of small wildlife against complicated backgrounds.

[https://www.sciencedirect.com/science/article/pii/S235293852600114X | Multiple authors | Remote Sensing Applications: Society and Environment | 2026-04]
 ''Deep learning for UAV thermal bird detection'' benchmarks dozens of YOLO variants for detecting little bustards from thermal drone imagery.

[https://www.sciencedirect.com/science/article/pii/S1574954126000683 | Multiple authors | Ecological Informatics | 2026-03]
 ''AEWD'' introduces a UAV benchmark containing more than 27,000 labeled examples of endangered Amur tigers, giant pandas, golden snub-nosed monkeys, and Sichuan takin.

[https://www.sciencedirect.com/science/article/pii/S1574954125005928 | Multiple authors | Ecological Informatics | 2026-02]
 A cloud-based deep-learning pipeline analyzes drone surveys from 163 seabird colonies containing more than 23,000 annotated birds.

[https://www.mdpi.com/1424-8220/26/3/790 | Chang Liu et al. | Sensors | 2026-01-24]
 ''YOLO-WL'' develops a lightweight wildlife-detection system intended for deployment aboard UAVs where computing capacity is limited.

[https://www.sciencedirect.com/science/article/pii/S0263224126023249 | Shaowen Wang et al. | Measurement | 2026]
 ''An algorithm for animal detection and counting in unmanned aerial imagery'' combines CNN and Transformer components to locate and count small wildlife targets in complex aerial scenes.

[https://www.frontiersin.org/journals/robotics-and-ai/articles/10.3389/frobt.2025.1695319/full | Multiple authors | Frontiers in Robotics and AI | 2026]
 ''WildDrone: autonomous drone technology for monitoring wildlife populations'' explores autonomous flight, animal tracking, population censuses, individual identification, posture estimation, and health monitoring.

[https://www.fisheries.noaa.gov/new-england-mid-atlantic/science-data/satellite-monitoring-north-atlantic-right-whales | NOAA Fisheries | NOAA Fisheries | 2026]
 NOAA's GAIA project combines very-high-resolution satellite imagery with artificial intelligence to locate critically endangered North Atlantic right whales from space.

[https://www.fisheries.noaa.gov/alaska/science-data/satellite-monitoring-cook-inlet-beluga-whales | NOAA Fisheries | NOAA Fisheries | 2026]
 Satellite imagery and geospatial AI are being developed to detect endangered Cook Inlet belugas despite highly turbid Alaskan coastal waters.

[https://www.sciencedirect.com/science/article/pii/S1574954125005047 | Multiple authors | Ecological Informatics | 2025-12]
 Drone imagery and Faster R-CNN detect, count, and distinguish blackbirds damaging crops by species, age, and sex, demonstrating AI applications to human-wildlife conflict.

[https://www.sciencedirect.com/science/article/pii/S1574954125004716 | Multiple authors | Ecological Informatics | 2025-12]
 Vision-language models classify salmon, trout, char, and other fishes from underwater video while temporal aggregation improves results across changing poses and water conditions.

[https://www.sciencedirect.com/science/article/pii/S1574954125002055 | Multiple authors | Ecological Informatics | 2025-12]
 A binocular-camera system performs real-time fish detection and body-size estimation on low-power hardware in Portuguese rivers affected by hydropeaking.

[https://www.sciencedirect.com/science/article/pii/S1574954125003899 | Multiple authors | Ecological Informatics | 2025-11]
 AI automatically identifies, counts, and maps common and little terns in crowded mixed-species breeding colonies where manual surveys are difficult and potentially disruptive.

[https://www.sciencedirect.com/science/article/pii/S1574954125003905 | Multiple authors | Ecological Informatics | 2025-11]
 YOLOv8 automatically detects freshwater eels in sonar imagery, creating a potential real-time monitoring tool for vulnerable migratory fishes.

[https://www.mdpi.com/2076-2615/15/21/3100 | Jun Liu et al. | Animals | 2025-10-25]
 A dual-model UAV system locates endangered spotted seals with lightweight onboard AI and then performs more precise identification after transmitting candidate images to a ground station.

[https://www.sciencedirect.com/science/article/pii/S1574954125001256 | Multiple authors | Ecological Informatics | 2025-07]
 An improved YOLOv8 and ByteTrack system identifies and counts tuna aboard fishing vessels using a model small enough for resource-constrained hardware.

[https://www.mdpi.com/2504-446X/9/7/455 | Nourdine Aliane | Drones | 2025-06-24]
 ''Drones and AI-Driven Solutions for Wildlife Monitoring'' reviews automatic species recognition, tracking, population estimation, habitat assessment, and anti-poaching applications.

[https://pubs.usgs.gov/publication/70272025 | Trevor Bak et al. | U.S. Geological Survey / Pacific Conservation Biology | 2025-06-23]
 Machine learning automatically detects Philippine deer and pigs in Guam camera videos and estimates their distance from cameras so population density can be calculated.

[https://www.sciencedirect.com/science/article/pii/S1574954125000226 | Multiple authors | Ecological Informatics | 2025-05]
 ''DeepFins'' combines spatial and temporal deep-learning information to detect both stationary and fast-moving fish in complex underwater videos.

[https://www.sciencedirect.com/science/article/pii/S1574954124005016 | Multiple authors | Ecological Informatics | 2025-03]
 YOLOv7 automates much of the underwater visual census process for Mediterranean fish communities and correctly identifies most species recorded during surveys.

[https://www.sciencedirect.com/science/article/pii/S1574954124005089 | Multiple authors | Ecological Informatics | 2025-03]
 ''FjordVision'' uses ROV video, YOLOv8, Mask R-CNN, and hierarchical classification to automate surveys of marine vegetation and fauna.

[https://www.sciencedirect.com/science/article/pii/S1574954124003844 | Multiple authors | Ecological Informatics | 2024-11]
 ''Collectively advancing deep learning for animal detection in drone imagery'' reviews the successes, limitations, datasets, and research gaps involved in automatically detecting animals from drones.

[https://www.nature.com/articles/s41598-024-68502-8 | Titus Venverloo and Fábio Duarte | Scientific Reports | 2024-08-12]
 ''Towards real-time monitoring of insect species populations'' uses millions of insect observations to develop computer-vision models capable of large-scale automated identification.

[https://www.sciencedirect.com/science/article/pii/S0048969724030249 | Multiple authors | Science of the Total Environment | 2024-07-20]
 Deep learning classifies 90 groups of mayflies, stoneflies, and caddisflies, potentially automating freshwater biomonitoring based on indicator insects.

[https://zslpublications.onlinelibrary.wiley.com/doi/full/10.1002/rse2.415 | Nathalie Pettorelli | Remote Sensing in Ecology and Conservation | 2024-06-17]
 ''Deep learning and satellite remote sensing for biodiversity monitoring and conservation'' reviews how neural networks can extract wildlife and habitat information from Earth-observation imagery.

[https://www.frontiersin.org/journals/conservation-science/articles/10.3389/fcosc.2024.1416706/full | R. L. Converse et al. | Frontiers in Conservation Science | 2024-06-05]
 ''Remote sensing and machine learning to improve aerial wildlife population surveys'' examines how AI can modernize traditional aircraft-based population counts.

[https://www.sciencedirect.com/science/article/pii/S1574954124000633 | Multiple authors | Ecological Informatics | 2024-05]
 ''STARdbi'' integrates automated insect detection, counting, body-size measurement, and image archiving into a scalable biodiversity-monitoring database.

[https://www.sciencedirect.com/science/article/pii/S1470160X24004734 | Adam Duarte et al. | Ecological Indicators | 2024-05]
 Passive acoustic monitoring combined with PNW-Cnet detects threatened marbled murrelets across Pacific Northwest forest landscapes and reconstructs seasonal patterns in calling activity.

[https://www.sciencedirect.com/science/article/pii/S1569843224000864 | Zeyu Xu et al. | International Journal of Applied Earth Observation and Geoinformation | 2024-04]
 ''A review of deep learning techniques for detecting animals in aerial and satellite images'' compares YOLO, Faster R-CNN, U-Net, ResNet, and related remote-sensing approaches.

[https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/1365-2664.14579 | Multiple authors | Journal of Applied Ecology | 2024-01-31]
 Passive acoustic sensors, BirdNET, and advanced statistical modeling reveal how a major Sierra Nevada wildfire changed bird-community composition and reduced biodiversity.

[https://www.sciencedirect.com/science/article/pii/S1574954124002218 | Multiple authors | Ecological Informatics | 2024]
 ''Will artificial intelligence revolutionize aerial surveys?'' reports a large-scale semi-automated African wildlife survey using oblique aerial photographs and deep learning.

[https://www.sciencedirect.com/science/article/abs/pii/S0924271620302409 | Multiple authors | ISPRS Journal of Photogrammetry and Remote Sensing | 2020-11]
 ''Wild animal survey using UAS imagery and deep learning'' applies a modified Faster R-CNN system to detecting kiang on the Tibetan Plateau.

[https://www.mdpi.com/1424-8220/16/1/97 | Luis F. Gonzalez et al. | Sensors | 2016-01-14]
 ''Unmanned Aerial Vehicles and Artificial Intelligence Revolutionizing Wildlife Monitoring and Conservation'' is an early influential paper on combining drones, imaging sensors, and AI for threatened-species surveys.

=== AI Bioacoustics and Automated Wildlife Listening ===

[https://www.sciencedirect.com/science/article/pii/S0006320726002879 | Multiple authors | Biological Conservation | 2026-09]
 ''BatSpot'' describes a retrainable neural network for recognizing bat echolocation, feeding buzzes, and social calls from acoustic monitoring data.

[https://www.sciencedirect.com/science/article/pii/S1574954126002128 | Multiple authors | Ecological Informatics | 2026-06]
 ''From croaks to species'' reviews machine learning, deep learning, and few-shot learning techniques for automatically identifying frogs and other anurans from their calls.

[https://www.sciencedirect.com/science/article/pii/S1574954126002220 | Multiple authors | Ecological Informatics | 2026-06]
 ''A noise-masking marine mammal sound classification method via multi-task learning'' develops AI designed to recognize marine mammals despite interfering underwater noise.

[https://www.sciencedirect.com/science/article/pii/S1574954126001391 | Multiple authors | Ecological Informatics | 2026-05]
 ''A review of acoustic systems for insect monitoring'' examines automated recognition of insects using sounds from calling, flight, movement, and other behaviors.

[https://www.sciencedirect.com/science/article/pii/S1574954126000634 | Multiple authors | Ecological Informatics | 2026-03]
 ''PNW-Cnet'' presents an evolving convolutional neural network capable of detecting more than 100 wildlife species and sound classes from large passive-acoustic datasets.

[https://www.sciencedirect.com/science/article/pii/S1574954126000506 | Multiple authors | Ecological Informatics | 2026-03]
 ''Promise and pitfalls'' evaluates BirdNET and VicFrogNET against more than 17,000 manually checked bird and frog detections, illustrating both the power and failure modes of acoustic AI.

[https://onlinelibrary.wiley.com/doi/10.1111/mms.70213 | Manuel Castellote et al. | Marine Mammal Science | 2026]
 ''Adaptive Acoustic Monitoring for Endangered Cook Inlet Beluga Whales in Complex Soundscapes'' uses deep learning, active learning, and audio-language models to improve whale monitoring.

[https://www.sciencedirect.com/science/article/pii/S1574954125003036 | Multiple authors | Ecological Informatics | 2025-12]
 ''Few-shot transfer learning enables robust acoustic monitoring of wildlife communities at the landscape scale'' adapts BirdNET to local communities using only a handful of examples per sound class.

[https://www.sciencedirect.com/science/article/pii/S1574954125004832 | Multiple authors | Ecological Informatics | 2025-12]
 ''Automated classification of albatross acoustic behaviour at sea'' uses convolutional neural networks to classify seabird sounds recorded by animal-borne devices.

[https://www.sciencedirect.com/science/article/pii/S1574954125002316 | Multiple authors | Ecological Informatics | 2025-12]
 ''Automated note annotation after bioacoustic classification'' combines machine learning and unsupervised clustering to improve detection of a cryptic owl.

[https://www.sciencedirect.com/science/article/pii/S1574954125002638 | Multiple authors | Ecological Informatics | 2025-12]
 A deep-learning bird-song detector developed using recordings from Doñana improves automated species identification and substantially reduces missed vocalizations relative to a general classifier.

[https://www.sciencedirect.com/science/article/pii/S1574954125004364 | Multiple authors | Ecological Informatics | 2025-12]
 Transfer learning successfully detects vocalizations of the critically endangered variable harlequin toad, demonstrating AI monitoring for extremely rare tropical amphibians.

[https://www.sciencedirect.com/science/article/pii/S1574954125002407 | Multiple authors | Ecological Informatics | 2025-12]
 A low-cost edge-computing network detects bird songs in remote areas and can transmit results even where ordinary 4G or 5G cellular coverage is unavailable.

[https://www.sciencedirect.com/science/article/pii/S1574954125002419 | Multiple authors | Ecological Informatics | 2025-12]
 Neural networks combining spectrogram representations and acoustic indices improve automated sound-based classification of birds and frogs.

[https://www.sciencedirect.com/science/article/pii/S1574954125002791 | Multiple authors | Ecological Informatics | 2025-12]
 ''ArcticSoundsNET'' adapts BirdNET embeddings to eight terabytes of Arctic recordings and substantially improves identification of wildlife sounds in northern Alaska.

[https://www.sciencedirect.com/science/article/pii/S1574954125003395 | Multiple authors | Ecological Informatics | 2025-12]
 ''Acoustic recognition of individuals in closed and open bird populations'' investigates whether AI can distinguish individual birds solely from their vocalizations.

[https://www.sciencedirect.com/science/article/pii/S1574954125003917 | Multiple authors | Ecological Informatics | 2025-11]
 ''First-of-its-kind AI model for bioacoustic detection using a lightweight associative memory Hopfield neural network'' investigates a computationally inexpensive and explainable alternative for passive-acoustic monitoring.

[https://www.sciencedirect.com/science/article/pii/S1574954125002043 | Multiple authors | Ecological Informatics | 2025-11]
 ''BattyCoda'' provides open-source software for machine-assisted annotation and classification of complex bat communication calls using relatively small training datasets.

[https://www.sciencedirect.com/science/article/pii/S157495412500175X | Multiple authors | Ecological Informatics | 2025-11]
 A two-stage machine-learning system filters false-positive acoustic detections of Ruffed Grouse, illustrating how secondary AI models can clean automated wildlife-monitoring results.

[https://www.sciencedirect.com/science/article/pii/S1574954125002900 | Multiple authors | Ecological Informatics | 2025-11]
 ''Territorial Acoustic Species Estimation'' uses networks of recorders and automated bird classifiers to estimate territorial species distributions across landscapes.

[https://www.sciencedirect.com/science/article/pii/S1574954125001815 | Multiple authors | Ecological Informatics | 2025-11]
 A free automated classifier analyzes hundreds of thousands of hours of recordings to track invasive cane toads across Australia.

[https://www.sciencedirect.com/science/article/pii/S1574954125001396 | Multiple authors | Ecological Informatics | 2025-11]
 Compressed neural networks detect North Atlantic right whale upcalls directly on edge hardware, potentially allowing conservation actions before acoustic data are returned to shore.

[https://www.sciencedirect.com/science/article/pii/S1574954125001116 | Multiple authors | Ecological Informatics | 2025-07]
 ''Continuous Real-Time Acoustic Monitoring of endangered bird species in Hawai‘i'' connects BirdNET models to an always-on field monitoring and alert system.

[https://www.sciencedirect.com/science/article/pii/S1574954125001311 | Multiple authors | Ecological Informatics | 2025-07]
 ''HawkEars'' builds a specialized Canadian avian classifier that outperforms broader acoustic models for many regional bird species.

[https://pmc.ncbi.nlm.nih.gov/articles/PMC12180400/ | Ciira wa Maina and Peter Njoroge | Philosophical Transactions of the Royal Society B | 2025-06-12]
 ''Comparing point counts, passive acoustic monitoring, citizen science and machine learning for bird species monitoring in the Mount Kenya ecosystem'' compares AI acoustics with traditional survey methods.

[https://www.sciencedirect.com/science/article/pii/S1574954125000123 | Multiple authors | Ecological Informatics | 2025-05]
 Machine-learning and Transformer systems detect substrate-borne insect vibrations, opening an automated monitoring pathway for insects largely missed by conventional microphones.

[https://www.sciencedirect.com/science/article/pii/S1574954124004485 | Multiple authors | Ecological Informatics | 2024-12]
 A deep-learning detector recognizes and classifies several marine mammal species from passive acoustic recordings collected around offshore wind developments.

[https://www.sciencedirect.com/science/article/pii/S1574954124004692 | Multiple authors | Ecological Informatics | 2024-12]
 Self-supervised learning and Transformer architectures improve recognition of montane bird vocalizations, including species poorly represented in training datasets.

[https://www.sciencedirect.com/science/article/pii/S1574954124004254 | Multiple authors | Ecological Informatics | 2024-12]
 An ensemble neural-network and anomaly-detection system automatically extracts deer vocalizations from long recordings and links acoustic activity with crop damage and fence breaches.

[https://www.sciencedirect.com/science/article/pii/S1574954124003935 | Multiple authors | Ecological Informatics | 2024-11]
 Data augmentation improves deep-learning classification of small-mammal vocalizations, particularly when only around 100 training examples exist for each species.

[https://www.sciencedirect.com/science/article/pii/S1574954124002528 | Multiple authors | Ecological Informatics | 2024-09]
 ''Leveraging transfer learning and active learning for data annotation in passive acoustic monitoring of wildlife'' investigates how AI can reduce the labor required to label enormous audio datasets.

[https://link.springer.com/article/10.1007/s42979-024-02859-5 | Harry Nel et al. | SN Computer Science | 2024-05-02]
 ''EcoSonicML'' applies machine learning to acoustic biodiversity monitoring in South African wetlands.

[https://www.sciencedirect.com/science/article/pii/S1574954124000591 | Multiple authors | Ecological Informatics | 2024-05]
 A deep-feature-loss network automatically removes environmental noise from bird recordings before species-recognition algorithms analyze them.

[https://www.sciencedirect.com/science/article/pii/S1574954124000700 | Multiple authors | Ecological Informatics | 2024]
 ''The bioacoustic soundscape of a pandemic'' uses continuous recording and deep learning to track changes in bird activity across consecutive years.

[https://www.sciencedirect.com/science/article/pii/S1574954123003205 | Multiple authors | Ecological Informatics | 2023-12]
 Convolutional neural networks automatically detect dolphin vocalizations and identify taxa, supporting monitoring of endangered Indian Ocean humpback dolphins.

[https://www.sciencedirect.com/science/article/pii/S1574954123002856 | Multiple authors | Ecological Informatics | 2023-11]
 ''Improving deep learning acoustic classifiers with contextual information for wildlife monitoring'' shows how geography and other metadata can substantially reduce false-positive bird and gibbon detections.

[https://www.sciencedirect.com/science/article/pii/S157495412300287X | Multiple authors | Ecological Informatics | 2023-11]
 ''Learning to detect an animal sound from five examples'' demonstrates few-shot systems capable of recognizing new wildlife sounds from extremely limited labeled data.

[https://www.sciencedirect.com/science/article/pii/S1574954123002972 | Multiple authors | Ecological Informatics | 2023-11]
 Machine learning and template matching automatically catalog oyster toadfish calls, providing information about an otherwise difficult-to-observe estuarine fish.

[https://www.sciencedirect.com/science/article/pii/S157495412300198X | Multiple authors | Ecological Informatics | 2023-11]
 Transfer learning adapts North Atlantic right whale detectors to unfamiliar acoustic environments without requiring researchers to build entirely new neural networks.

[https://www.sciencedirect.com/science/article/pii/S1574954123002625 | Multiple authors | Ecological Informatics | 2023-11]
 Active learning reduces the annotation effort needed to construct an automated call recognizer for the rare southern black-throated finch.

[https://www.sciencedirect.com/science/article/pii/S1574954123000948 | Multiple authors | Ecological Informatics | 2023-07]
 Citizen-science recordings and convolutional neural networks are combined to recognize dozens of bird species while examining how human and environmental noise affect accuracy.

[https://www.sciencedirect.com/science/article/pii/S1574954123000249 | Multiple authors | Ecological Informatics | 2023-05]
 BirdNET's internal feature embeddings are used to move beyond simple species identification and investigate different types of calls within individual species.

[https://www.mdpi.com/2071-1050/15/9/7128 | Multiple authors | Sustainability | 2023]
 ''A Methodological Literature Review of Acoustic Wildlife Monitoring Using Artificial Intelligence Tools and Techniques'' reviews dozens of AI bioacoustic studies involving birds, mammals, and other wildlife.

[https://www.nature.com/articles/s41467-023-41693-w | Multiple authors | Nature Communications | 2023]
 ''Soundscapes and deep learning enable tracking biodiversity recovery in tropical forests'' demonstrates that automated acoustic measurements can reveal biological recovery following tropical forest restoration.

[https://besjournals.onlinelibrary.wiley.com/doi/10.1111/1365-2664.14280 | Ali Khalighifar et al. | Journal of Applied Ecology | 2022-08-23]
 ''NABat ML'' uses deep learning and hundreds of thousands of bat-call spectrograms to support automated North American bat population monitoring.

[https://www.sciencedirect.com/science/article/pii/S1574954121000273 | Stefan Kahl et al. | Ecological Informatics | 2021-03]
 ''BirdNET: A deep learning solution for avian diversity monitoring'' presents the influential neural-network system that made large-scale automatic identification of birds from sound practical.

=== Individual Recognition, Behavior and Human-Wildlife Conflict ===

[https://www.sciencedirect.com/science/article/pii/S2352938526000571 | Multiple authors | Remote Sensing Applications: Society and Environment | 2026-01]
 ''Smart surveillance with conversational alerts for wild elephant early warning'' integrates YOLO detection, CCTV cameras, GPS information, and automated messaging for community warnings.

[https://doi.org/10.1016/j.inffus.2026.104323 | Multiple authors | Information Fusion | 2026]
 ''From species-specific models to universal re-ID'' reviews recent progress toward AI systems capable of identifying individuals across many animal species.

[https://wildlabs.net/sites/default/files/2026-01/Chen%20et%20al.%20-%202025%20-%20PantherAI%20An%20autonomous%20behavioural%20monitoring%20tool%20for%20assessing%20activity%20budget%20and%20space%20use.pdf | Li-Dunn Chen et al. | Ecological Informatics | 2026]
 ''PantherAI'' uses automated computer vision to monitor activity patterns and habitat use of a zoo-housed tiger, illustrating applications to continuous behavioral monitoring.

[https://www.sciencedirect.com/science/article/pii/S0960982225016707 | Multiple authors | Current Biology | 2026]
 Pose-aware metric learning distinguishes individual Alaskan coastal brown bears despite dramatic seasonal changes in body mass, fur, viewing angle, and appearance.

[https://spotid.org/ | SpotID Project | SpotID | 2026]
 SpotID uses deep learning and individually distinctive coat patterns to suggest matches between leopard photographs collected by camera traps.

[https://www.sciencedirect.com/science/article/pii/S1574954125004285 | Multiple authors | Ecological Informatics | 2025-12]
 Synchronized cameras, tracking algorithms, and three-dimensional mapping continuously identify swimming birds and even classify eye state for behavioral and sleep studies.

[https://www.sciencedirect.com/science/article/pii/S1617138125002407 | Multiple authors | Journal for Nature Conservation | 2025-12]
 ''Deep-track'' uses real-time video and deep learning to detect elephants, deer, leopards, and other wildlife approaching settlements in human-wildlife conflict zones.

[https://www.sciencedirect.com/science/article/pii/S1574954125003887 | Multiple authors | Ecological Informatics | 2025-11]
 A deep-learning system tracks and identifies individual white-tailed eagles in the wild with greater than 93 percent recognition accuracy.

[https://www.sciencedirect.com/science/article/pii/S1574954125001153 | Multiple authors | Ecological Informatics | 2025-07]
 ''BEHAVE'' is an open-source tool that combines AI animal detections with behavioral coding so researchers can rapidly analyze large wildlife-video datasets.

[https://www.jas.shu.edu.cn/EN/10.3969/j.issn.0255-8297.2025.02.012 | Yang Wan et al. | Journal of Applied Sciences | 2025-04-03]
 An improved EfficientNet architecture combines routing attention and self-calibrated convolution to recognize individual leopards in difficult infrared camera-trap images.

[https://www.sciencedirect.com/science/article/pii/S1574954124003893 | Multiple authors | Ecological Informatics | 2025-03]
 DeepLabCut and neural networks quantify facial movements of lemurs and small apes and automatically distinguish facial gestures associated with vocalizing animals.

[https://www.sciencedirect.com/science/article/pii/S1574954124004977 | Multiple authors | Ecological Informatics | 2025-03]
 ''Mamba-MSQNet'' recognizes multiple animal behaviors while substantially reducing computation compared with Transformer-based video models.

[https://conbio.onlinelibrary.wiley.com/doi/10.1111/csp2.70186 | Multiple authors | Conservation Science and Practice | 2025]
 An AI-embedded camera-alert system in West Bengal detected elephants approaching communities and transmitted near-real-time warnings to field teams responding to human–elephant conflict.

[https://mountainscholar.org/items/90ff47cc-0a8a-4b1e-9f76-34972ac2ac20 | Cheng Guo | Colorado State University | 2025]
 Research on African leopards develops automated and human-in-the-loop methods for clustering unlabeled camera-trap photographs into individual animals.

[https://pubmed.ncbi.nlm.nih.gov/38244194/ | Multiple authors | American Journal of Primatology | 2024]
 Passive acoustic monitoring and deep learning automatically detect black-and-white ruffed lemurs in Madagascar and outperform conventional in-person surveys in several respects.

[https://www.sciencedirect.com/science/article/pii/S1574954123004016 | Multiple authors | Ecological Informatics | 2023-12]
 Neural networks identify both regional song dialects and individual globally threatened yellow cardinals from audio recordings.

[https://www.sciencedirect.com/science/article/pii/S1574954123002431 | Multiple authors | Ecological Informatics | 2023-11]
 Two automated identification algorithms are compared on a large snow-leopard image collection, advancing computer-assisted capture-recapture analysis.

[https://www.sciencedirect.com/science/article/pii/S1574954123000651 | Multiple authors | Ecological Informatics | 2023-07]
 Siamese neural networks re-identify individual undulate skates from photographs, offering a non-invasive population-monitoring method for a marine species of conservation concern.

[https://www.sciencedirect.com/science/article/pii/S1574954123000237 | Multiple authors | Ecological Informatics | 2023-05]
 ''An experiment on animal re-identification from video'' compares machine-learning approaches for repeatedly recognizing individual fish, pigeons, and pigs.

[https://academic.oup.com/bioscience/article/73/10/748/7261057 | Multiple authors | BioScience | 2023]
 ''Mitigating human–wildlife conflict and monitoring endangered tigers using a real-time camera-based alert system'' describes TrailGuard AI cameras capable of detecting tigers and suspected poachers and transmitting alerts in roughly seconds.

[https://pmc.ncbi.nlm.nih.gov/articles/PMC10645515/ | Multiple authors | Conservation Science and Practice / review | 2023]
 ''Elephants and algorithms'' reviews AI applications using cameras, microphones, geophones, drones, satellites, and other sensors for elephant conservation.

[https://www.frontiersin.org/journals/marine-science/articles/10.3389/fmars.2023.1099479/full | Multiple authors | Frontiers in Marine Science | 2023]
 ''Scaling whale monitoring using deep learning'' demonstrates a human-in-the-loop system that rapidly analyzes aerial photographs for beluga whale population surveys.

[https://pmc.ncbi.nlm.nih.gov/articles/PMC9265132/ | Multiple authors | Animals | 2022]
 ''The Future of Artificial Intelligence in Monitoring Animal Identification, Health, and Behaviour'' discusses continuous AI monitoring of animal identity, movement, health, posture, and behavior.

[https://www.sciencedirect.com/science/article/pii/S1574013720303890 | Prashanth C. Ravoor and Sudarshan T.S.B. | Computer Science Review | 2020-11]
 ''Deep Learning Methods for Multi-Species Animal Re-identification and Tracking'' surveys AI techniques designed to recognize individual animals as they move between cameras.

[https://pmc.ncbi.nlm.nih.gov/articles/PMC7415925/ | Daniel Schofield et al. | iScience | 2020]
 ''Automatic Identification of Individual Primates with Deep Learning Techniques'' demonstrates AI facial recognition across more than 1,000 individual primates and several carnivore species.

=== Applied Conservation AI, Anti-Poaching and Monitoring Platforms ===

[https://bds.org.uk/2026/06/17/how-ai-could-revolutionise-deer-monitoring/ | British Deer Society | British Deer Society | 2026-06-17]
 ''How AI Could Revolutionise Deer Monitoring'' provides a practitioner-oriented assessment of computer vision, automated data processing, practical limitations, and opportunities for deer management.

[https://blog.google/company-news/outreach-and-initiatives/sustainability/speciesnet-open-source-ai-wildlife/ | Tanya Birch and Dan Morris | Google | 2026-03-06]
 ''How our open-source AI model SpeciesNet is helping to promote wildlife conservation'' describes worldwide use of SpeciesNet to identify thousands of wildlife categories in camera-trap photographs.

[https://www.sciencedirect.com/science/article/pii/S3050607726000024 | Multiple authors | Human Settlements and Sustainability | 2026-03]
 A proposed anti-poaching framework combines wildlife identification, behavior recognition, animal-sound classification, and UAV video analysis to identify potential threats inside protected areas.

[https://www.nature.com/articles/d44151-026-00027-7 | Nature India | Nature India | 2026-02-11]
 ''Turning camera traps into conservation tools'' highlights a large annotated dataset designed to train AI for monitoring vulnerable houbara bustards and potential nest predators.

[https://sentinel.conservationxlabs.com/ | Conservation X Labs | Sentinel | 2026]
 Sentinel combines field hardware, local AI models, remote communications, and dashboards to provide near-real-time wildlife detections in remote conservation landscapes.

[https://www.fisheries.noaa.gov/new-england-mid-atlantic/science-data/geospatial-artificial-intelligence-animals | NOAA Fisheries | NOAA Fisheries | 2026]
 ''Geospatial Artificial Intelligence for Animals'' is building a cloud platform that uses satellite imagery and AI to locate whales and eventually other wildlife over enormous and inaccessible areas.

[https://iucn.org/story/202512/tech4nature-2025-harnessing-technology-deliver-conservation-impact-scale | IUCN | International Union for Conservation of Nature | 2025-12-18]
 IUCN's Tech4Nature program reports AI-enabled jaguar monitoring, smart camera traps for black-bear conflict, acoustic monitoring of bats, and integrated sensor networks across protected areas.

[https://wildlabs.net/event/ai-conservation-edge-project-sparrow | WILDLABS / Microsoft AI for Good | WILDLABS | 2025-12-01]
 Project SPARROW brings solar-powered edge computing to remote conservation sites so camera photographs and acoustic recordings can be analyzed locally using AI.

[https://news.mongabay.com/2025/11/what-central-parks-squirrel-census-says-about-conservation-tech-interview-with-okalas-robin-whytock/ | Mongabay | Mongabay | 2025-11]
 Conservation technologist Robin Whytock discusses applying global AI camera-trap and bioacoustic models to large datasets involving chimpanzees, squirrels, and other wildlife.

[https://wildlabs.net/event/2025-workshop-camera-traps-ai-and-ecology | Allen Institute for AI / WILDLABS | WILDLABS | 2025-09-09]
 The Camera Traps, AI, and Ecology workshop brings together ecologists, protected-area managers, conservation technologists, and AI researchers working on automated wildlife monitoring.

[https://news.mongabay.com/2025/09/turning-camera-traps-into-real-time-sentinels-interview-with-conservation-x-labs-dante-wasmuht/ | Abhishyant Kidangoor | Mongabay | 2025-09]
 Conservation X Labs' Sentinel attaches edge AI to ordinary camera traps so invasive species, endangered wildlife, and potential threats can be reported without waiting months for memory cards to be retrieved.

[https://www.usgs.gov/mission-areas/ecosystems/science/artificial-intelligence-usgs-ecosystems-mission-area-fish-and | U.S. Geological Survey | USGS | 2025-06-23]
 USGS summarizes AI programs that process wildlife images, video, and sound for species identification, population monitoring, animal behavior research, disease detection, and aerial surveys.

[https://wildlabs.net/en/discussion/wildlabs-awards-2025-trapper-keeper-scalable-and-energy-efficient-open-source-camera | Izabela Stachowicz | WILDLABS | 2025-05-30]
 ''Trapper Keeper'' is an open-source AI-powered infrastructure project intended to automate species identification and improve large-scale camera-trap data management.

[https://www.worldwildlife.org/news/stories/how-thermal-cameras-and-ai-are-powering-rhino-conservation-success-in-kenya/ | Whitney Kent | WWF | 2025-05-14]
 ''How thermal cameras and AI are powering rhino conservation success in Kenya'' describes AI-equipped night-vision systems that detect people, animals, and vehicles and rapidly alert anti-poaching teams.

[https://www.australianwildlife.org/news-and-resources/wildlife-matters/smart-sanctuaries | Alana Burton and Yezmin Assad | Australian Wildlife Conservancy | 2025-05-02]
 Australian Wildlife Conservancy describes its internally developed AI Species Classifier for quickly processing camera-trap records of native and invasive animals.

[https://www.worldwildlife.org/news/stories/using-the-power-of-ai-to-identify-and-track-species/ | Abby Hehmeyer | WWF | 2025-03-03]
 ''Using the power of AI to identify and track species'' explains how open-source SpeciesNet processes camera-trap imagery and supports jaguar, primate, and other wildlife monitoring.

[https://blog.google/company-news/outreach-and-initiatives/entrepreneurs/ai-nature-climate-accelerator-nonprofits-speciesnet/ | Mike Werner | Google | 2025-03-03]
 ''3 new ways we're working to protect and restore nature using AI'' covers SpeciesNet's open-source release and broader programs supporting AI-based nature conservation.

[https://wildlabs.net/discussion/we-are-releasing-speciesnet | Ștefan Istrate | WILDLABS | 2025-03-03]
 Conservation technologists discuss the open-source release of SpeciesNet, a global camera-trap classifier capable of recognizing more than 2,000 wildlife species.

[https://www.fisheries.noaa.gov/endangered-species-conservation/advanced-sampling-and-technology-extinction-risk-reduction-and | NOAA Fisheries | NOAA Fisheries | 2025]
 ''Advancing Technologies for Protected Species Conservation'' discusses machine learning for detecting whales and other protected wildlife in acoustic recordings and imagery.

[https://www.frontiersin.org/journals/ecology-and-evolution/articles/10.3389/fevo.2025.1527976/full | Multiple authors | Frontiers in Ecology and Evolution | 2025]
 ''The rising tide of conservation technology'' examines AI-enabled surveillance, connected cameras, real-time wildlife-crime alerts, and anti-poaching systems.

[https://wildlabs.net/en/article/announcing-wildlabs-awards-2025-grantees | WILDLABS | WILDLABS | 2025]
 WILDLABS' conservation-technology grants include AI camera infrastructure, automated species recognition, human-wildlife conflict tools, and scalable insect-monitoring technologies.

[https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.70169 | Multiple authors | Methods in Ecology and Evolution | 2025]
 ''Sensors versus surveyors'' compares passive acoustic monitoring assisted by BirdNET embeddings with camera traps and conventional observers for detecting terrestrial mammals across eastern Australia.

[https://www.worldwildlife.org/news/stories/how-artificial-intelligence-buys-valuable-time-to-protect-wildlife/ | Abby Hehmeyer | WWF | 2024-05-14]
 ''How artificial intelligence buys valuable time to protect wildlife'' describes processing more than seven million Australian post-fire camera-trap images using Wildlife Insights.

[https://news.mongabay.com/2024/03/gundi-tool-gets-conservation-hardware-and-software-talking-the-same-language/ | Abhishyant Kidangoor | Mongabay | 2024-03-12]
 ''Gundi'' links camera traps, wildlife trackers, acoustic recorders, and conservation software so information from different sensor systems can flow into common monitoring platforms.

[https://www.worldwildlife.org/our-work/oceans/mangroves/manglaria-using-artificial-intelligence-to-save-mangroves-in-a-changing-climate/ | WWF | WWF | 2023]
 ''ManglarIA'' combines weather stations, drones, camera traps, ecological sensors, and AI to understand mangrove health and associated wildlife.

[https://www.microsoft.com/en-us/research/articles/biodiversity-surveys-research-compilation/ | Microsoft Research | Microsoft | 2022-03-18]
 ''Accelerating Biodiversity Surveys with AI'' compiles Microsoft work involving camera traps, aerial imagery, microphones, active learning, and automated wildlife detection.

[https://techcommunity.microsoft.com/blog/iotblog/wildlife-monitoring-and-conservation-with-azure-percept/3390910 | Microsoft / BearID Project | Microsoft | 2022]
 ''Wildlife Monitoring and Conservation with Azure Percept'' demonstrates edge-computing technology capable of detecting and individually identifying bears near the camera rather than waiting for data retrieval.

[https://blog.google/products/earth/ai-finds-where-the-wild-things-are/ | Google Earth Outreach | Google | 2019]
 ''Using AI to find where the wild things are'' introduces Wildlife Insights and describes how cloud AI can process camera-trap imagery many times faster than manual classification.

[https://www.worldwildlife.org/our-work/science/artificial-intelligence-and-conservation/ | WWF | WWF | n.d.]
 ''Artificial intelligence and conservation'' provides an overview of WWF applications involving camera traps, acoustic recordings, thermal systems, wildlife crime, deforestation, and ecosystem monitoring.