Abstract:
Understanding the emotional states of pets is a significant aspect of their well-being and effective
communication between humans and animals. This research investigates the development of a deep
learning-based system for facial expression recognition in pets. Leveraging convolutional neural
networks (CNNs) and transfer learning techniques, the proposed model aims to accurately detect and
classify diverse facial expressions exhibited by various animal species, such as dogs, cats, and others.
The study involves the collection and curation of a comprehensive dataset comprising annotated images
of pets displaying different emotional cues such as angry, happy, sad, and other. Preprocessing methods
tailored to account for the variability in animal faces are employed to enhance model robustness and
generalization. Through extensive experimentation and evaluation, the effectiveness and reliability of the
developed framework in recognizing and interpreting pet facial expressions are assessed. EfficientNetB5
is used for transfer learning and the accuracy of the detection is around 87%. The outcomes of this
research pave the way for innovative applications in veterinary care, animal behavior analysis, and
human-animal interaction, fostering a deeper understanding of pet emotions and improving the quality
of relationships between pets and their human companions.
