Fusion of Deep Learning Features for Wild Animal Detection

Dhillon, Anamika; Verma, Gyanendra K.

doi:10.1007/978-3-030-85365-5_2

Cited by 2 publications

(1 citation statement)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The application of machine learning models, especially convolutional neural networks, can save a significant amount of time for researchers by automating the extraction of information from images (Norouzzadeh et al, 2018). Furthermore, data extraction from photographic images through AI contributes to the reduction in costs associated with information collection, making projects more accessible to those without large teams of human volunteers, enabling more cost-effective and, in some cases, real-time monitoring, and the productivity gains and time reduction provided by the integration of AI can be positive factors, expediting studies related to wildlife documentation and conservation (Norouzzadeh et al, 2018;Curran et al, 2022;Dhillon and Verma, 2022).…”

Section: The Effect Of Images On People's Perception Of Species and T...mentioning

confidence: 99%

Revisiting the debate: documenting biodiversity in the age of digital and artificially generated images

Campos,

Oliveira,

Vieira

et al. 2023

Web Ecol.

View full text Add to dashboard Cite

Abstract. This study examines the risks associated with relying solely on images for documenting new species records, rediscoveries, taxonomic descriptions, and distribution expansions. We highlight concerns regarding image authenticity, especially in cases where images may be altered, adulterated, or AI (artificial intelligence)-generated, potentially leading to inaccuracies in biodiversity documentation. To illustrate the evolving challenges, we conducted an experiment with 621 participants who assessed nine AI-generated images. Surprisingly, six were deemed authentic, while three raised doubts, highlighting the difficulty in discerning AI-generated content. Our main message emphasizes the critical role of trust in biodiversity documentation, particularly for taxonomy and conservation, and how eroded trust can hinder conservation efforts. Improved communication and collaboration between taxonomists and conservationists are needed, emphasizing scientific integrity. We urge a reevaluation of journal policies concerning data validation, especially in articles relying on images as primary evidence, to preserve the credibility of scientific research amidst technological advancements.

show abstract