The ‘mice_of’ dataset was used to evaluate performance for tracking mice under optimal imaging conditions (high contrast) and with variable numbers of animals. The dataset consisted of videos from C57BL/6J male (n = 17) and female (n = 20) mice acquired from Jackson Laboratory (RRID:IMSR_JAX:000664, Jackson Laboratory). Groups of four and five mice were formed from same-sex littermates, and groups of two same-sex mice were picked randomly from different litters and interacted with each other in the open field for the first time. During video recording, mice moved freely in a 45.7 × 45.7-cm open-field arena with a clear acrylic floor. Videos were captured from below with infrared illumination using a Point Grey Blackfly S camera at a resolution of 1.97 pixels per mm at 80 FPS.
Experimental procedures were approved by the Princeton University Institutional Animal Care and Use Committee and conducted in accordance with the National Institutes of Health guidelines for the humane care and use of laboratory animals. Mice used in this study had at least 1 week of acclimation to the Princeton Neuroscience Institute vivarium in group cages with food and water ad libitum under a reversed 12-h–12-h dark–light cycle (light, 19:30–07:30) and were habituated in the dark test room for at least 30 min before experimental procedures were performed.
For this dataset, we labeled 1,000 frames (2,950 instances) with a skeleton consisting of 11 nodes: nose, neck, ‘L_ear’, ‘R_ear’, ‘L_Fr_paw’, ‘R_Fr_paw’, ‘tail_base’, ‘L_Hi_paw’, ‘R_Hi_paw’, ‘tail_mid’ and ‘tail_end’; and ten edges: neck to ‘L_Fr_paw’, neck to ‘R_Fr_paw’, ‘tail_base’ to ‘R_Hi_paw’, ‘tail_base’ to ‘L_Hi_paw’, ‘tail_base’ to ‘tail_mid’, ‘tail_mid’ to ‘tail_end’, neck to nose, neck to ‘R_ear’, neck to ‘L_ear’ and ‘tail_base’ to neck. Labels were randomly split into 800 training, 100 validation and 100 test frames.
Experimental procedures were approved by the Princeton University Institutional Animal Care and Use Committee and conducted in accordance with the National Institutes of Health guidelines for the humane care and use of laboratory animals. Mice used in this study had at least 1 week of acclimation to the Princeton Neuroscience Institute vivarium in group cages with food and water ad libitum under a reversed 12-h–12-h dark–light cycle (light, 19:30–07:30) and were habituated in the dark test room for at least 30 min before experimental procedures were performed.
For this dataset, we labeled 1,000 frames (2,950 instances) with a skeleton consisting of 11 nodes: nose, neck, ‘L_ear’, ‘R_ear’, ‘L_Fr_paw’, ‘R_Fr_paw’, ‘tail_base’, ‘L_Hi_paw’, ‘R_Hi_paw’, ‘tail_mid’ and ‘tail_end’; and ten edges: neck to ‘L_Fr_paw’, neck to ‘R_Fr_paw’, ‘tail_base’ to ‘R_Hi_paw’, ‘tail_base’ to ‘L_Hi_paw’, ‘tail_base’ to ‘tail_mid’, ‘tail_mid’ to ‘tail_end’, neck to nose, neck to ‘R_ear’, neck to ‘L_ear’ and ‘tail_base’ to neck. Labels were randomly split into 800 training, 100 validation and 100 test frames.
Free full text: Click here
