A computer-vision dataset of three common marmosets (Callithrix jacchus) living
together in a single cage, recorded by two cameras that face each other. Every
marmoset in every annotated frame carries three labels at once:
an instance segmentation polygon that outlines the visible body.
a tracking identity that is consistent across frames within one video.
a posture label, one of five classes: climb, lay, sit, stand, or walk/jump.
3… See the full description on the dataset page:
https://huggingface.co/datasets/makuda/MC-Marmo.
