Researchers collected about 127 audio-visual dog recordings in Mangalore to build a multimodal canine emotion dataset covering Happy, Sad, Angry, and Relaxed
This work collected approximately 127 audio-visual dog recordings from various localities of Mangalore, organized them into four emotion classes—Happy, Sad, Angry, and Relaxed—and identified and assigned emotion labels through clustering algorithms, aiming to provide a data basis for multimodal, audio-based, video-based, and emotion classification systems and to investigate the feasibility of recognizing dog emotions from audio-visual cues.