Decoding Animal Emotions: Predicting Reactions with Deep Learning for Enhanced Understanding

Dogs, integral members of our environment, warrant our attention and understanding of their emotions, including anger, fear, happiness, and relaxation. This study employs deep learning models for pattern recognition to predict canine emotions dynamically through poses. A dataset of 574 dog images capturing various angles and postures is utilized, comprising 149 samples for anger, 143 for fear, 142 for happiness, and 140 for relaxation. Convolutional Neural Networks (CNNs), including VGG16, EfficientNetB0, and InceptionV3, are employed, along with vision transformer, to assess their efficacy in emotion prediction. Notably, EfficientNetB0 demonstrated outstanding performance, achieving an accuracy of 80±5