EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Animal affective-state analysis using Convolutional Neural Networks: Methods, datasets, applications, challenges, and future directions

Biswajit Mandal, Dilip Kumar

Letters in Animal Biology · 10.11.2026 · 6 (1) · 193-206

DOI: 10.62310/liab.v6i1.454 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Letters in Animal Biology, online erschienen am 10.11.2026. Verfasst von Biswajit Mandal und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Animal emotions are inner affective and bodily states that make certain physiological, behavioral, and survival responses to environmental changes. Comprehension of the animal behavior and emotional states is pivotal a step to improve human-animal relationship and enhance animal welfare. Convolutional Neural Networks (CNNs) are among the popular methodologies to analyze the animal-emotion related data. However, CNNs face challenges such as dataset scarcity, inter-breed morphological variability, and challenges in associating subtle behavioral micro expressions with emotion labels. CNNs help predict animal emotions via successive convolutional layers, images pass through these layers and automatically extract features like eyebrow, mouth curvature, and classify them into different emotional states. Various methods, including AlexNet, ResNet, EfficientNet, MobileNet, have been commonly used for feature extraction and classification. Leveraging the potential of CNN in image analysis and data generation, it can be applied to animal emotion-state analysis. This review outlines the fundamental issues of emotion classification and information retrieval by CNN techniques; and critically evaluates the available literature on the CNN applications in animal affective-state analysis, advocates research gaps, and proposes future research direction. Combining CNNs with data augmentation technique can greatly enhanced model accuracy and prediction, and addresses the issues with small training datasets. Various methods for emotion detection and very few available datasets used by researchers or to used by other researchers for future research are there. The use of CNN-based approaches will greatly benefit the researchers to enhance their studies and help them develop and implement real-world applications.

Zitieren

APA

Mandal, B., & Kumar, D. (2026). Animal affective-state analysis using Convolutional Neural Networks: Methods, datasets, applications, challenges, and future directions. Letters in Animal Biology, 6(1), 193-206. https://doi.org/10.62310/liab.v6i1.454

Vancouver

Mandal B, Kumar D. Animal affective-state analysis using Convolutional Neural Networks: Methods, datasets, applications, challenges, and future directions. Letters in Animal Biology. 2026;6(1):193-206. doi:10.62310/liab.v6i1.454

BibTeX
@article{Mandal2026,
  author = {Mandal, Biswajit and Kumar, Dilip},
  title = {Animal affective-state analysis using Convolutional Neural Networks: Methods, datasets, applications, challenges, and future directions},
  journal = {Letters in Animal Biology},
  year = {2026},
  volume = {6},
  number = {1},
  pages = {193--206},
  doi = {10.62310/liab.v6i1.454}
}
Kostenlos starten →