Identification of animals and recognition of their actions in wildlife videos using deep learning techniques

Schindler, Frank; Steinhage, Volker

doi:10.1016/j.ecoinf.2021.101215

Cited by 40 publications

(14 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…), thus reducing the generalization capabilities to new sensor deployments or regions. For these reasons, detecting and classifying animal species with DL for the purpose of population estimates is becoming increasingly popular for images 52,53 , acoustic spectrograms 95 , and videos 96 . Models performing accurately and robustly on specific classes (e.g., the MegaDetector -see Box 2 -or AIDE; see Table 1) are now being used routinely and integrated within open systems supporting ecologists performing both labeling and detection, respectively counting of their image databases.…”

Section: Machine Learning To Scale-up and Automate Animal Ecology And...mentioning

confidence: 99%

Perspectives in machine learning for wildlife conservation

et al. 2022

View full text Add to dashboard Cite

Inexpensive and accessible sensors are accelerating data acquisition in animal ecology. These technologies hold great potential for large-scale ecological understanding, but are limited by current processing approaches which inefficiently distill data into relevant information. We argue that animal ecologists can capitalize on large datasets generated by modern sensors by combining machine learning approaches with domain knowledge. Incorporating machine learning into ecological workflows could improve inputs for ecological models and lead to integrated hybrid modeling tools. This approach will require close interdisciplinary collaboration to ensure the quality of novel approaches and train a new generation of data scientists in ecology and conservation.

show abstract

Section: Machine Learning To Scale-up and Automate Animal Ecology And...mentioning

confidence: 99%

Perspectives in machine learning for wildlife conservation

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Especially the concept of transfer learning, using models trained for a specific image processing task to perform a related task, potentially even in a different scientific field, is promising for the study of ecological change, where labelled image data for model training are currently still scarce (e.g. Heredia, 2017; Schindler & Steinhage, 2021; Younis et al, 2020). Using this concept, models trained on a collection of freely available labelled images from trivial subjects such as cats, cars or houses [e.g.…”

Section: Three Promising Avenues For the Use Of Photos In Ecological ...mentioning

confidence: 99%

The use of photos to investigate ecological change

et al. 2022

View full text Add to dashboard Cite

Global change is causing ecosystems to change at unprecedented rates and the urgency to quantify ecological change is high. We therefore need all possible sources of ecological data to address key knowledge gaps. Ground‐based photos are a form of remote sensing and an unconventional data source with a high potential to improve our understanding of ecological change. They can provide invaluable information on ecological conditions in the past and present at relevant spatiotemporal scales that is very difficult to obtain with other approaches. Here we review the use of ground‐based photos in a set of relevant ecological research topics, such as biodiversity and community ecology, phenology, global change ecology and landscape ecology. We highlight three main photo‐based methods in ecological research (repeat photography, time‐lapse photography and public archives), alongside which we discuss three case studies to demonstrate novel applications of these methods, to answer fundamental ecological questions. Synthesis. Photos can significantly support ecological research to improve our understanding of biotic responses in a rapidly changing world. Photos cover relatively large temporal and spatial scales, and can provide large amounts of information with limited time investment. To exploit their full potential, we need to invest not only in technological advances to compile, process and analyse images but also in proper data management.

show abstract

“…In the past two decades, facial recognition technology has developed, allowing for unobtrusive individual recognition and tracking [16,21,37,38]. With recent advancements in technology, computer software involving machine and deep learning, as well as artificial neural networks (ANNs), have been trained to provide surveillance of animals with regard to their location, species identity, individual identity, and/or behaviour (e.g., [19,21,27,30,31,[39][40][41][42][43][44]). Machine learning focuses on algorithms and data to imitate the way humans learn; deep learning is a subfield of machine learning.…”

Section: Technology Typesmentioning

confidence: 99%

The Future of Artificial Intelligence in Monitoring Animal Identification, Health, and Behaviour

Congdon

Hosseini

Gading

et al. 2022

Animals

View full text Add to dashboard Cite

With many advancements, technologies are now capable of recording non-human animals’ location, heart rate, and movement, often using a device that is physically attached to the monitored animals. However, to our knowledge, there is currently no technology that is able to do this unobtrusively and non-invasively. Here, we review the history of technology for use with animals, recent technological advancements, current limitations, and a brief introduction to our proposed novel software. Canadian tech mogul EAIGLE Inc. has developed an artificial intelligence (AI) software solution capable of determining where people and assets are within public places or attractions for operational intelligence, security, and health and safety applications. The solution also monitors individual temperatures to reduce the potential spread of COVID-19. This technology has been adapted for use at the Toronto Zoo, initiated with a focus on Sumatran orangutans (Pongo abelii) given the close physical similarity between orangutans and humans as great ape species. This technology will be capable of mass data collection, individual identification, pose estimation, behaviour monitoring and tracking orangutans’ locations, in real time on a 24/7 basis, benefitting both zookeepers and researchers looking to review this information.

show abstract

Identification of animals and recognition of their actions in wildlife videos using deep learning techniques

Cited by 40 publications

References 12 publications

Perspectives in machine learning for wildlife conservation

Perspectives in machine learning for wildlife conservation

The use of photos to investigate ecological change

The Future of Artificial Intelligence in Monitoring Animal Identification, Health, and Behaviour

Contact Info

Product

Resources

About