Table 1.
Main properties of publicly available datasets. N stands for Newborns, I for Infants and T for Toddlers, Misc for Miscellaneous, NA for Not Applicable, U for Unknown.
| Database | Contents | Frame Size |
Age Range |
Info | Frames | Labels |
|---|---|---|---|---|---|---|
| BabyPose [37] | 16 Videos | 640 × 480 | N | Depth 8 bit/16 bit |
16,000 | 12 Body Landmarks |
| MINI-RGBD [38] | 12 Videos | 640 × 480 | I | RGB/D | 12,000 | 25 Body Landmarks |
| SyRIP [41] | Images | Misc | I | RGB | 2000 | 17 Body Landmarks |
| Dataset [42] | 85 Youtube Video URLs |
Misc | I | RGB | NA | 18 Body Landmarks |
| SSBD [44] | 75 Youtube Video URLs |
Misc | NA | RGB | U | Behaviors |
| MMDB [45] | 160 Videos | Misc | T | Multimodal | U | ASD Diagnosis |
| Tariq [46] | 162 Videos | Misc | T | RGB | U | Behaviors |
| DREAM [47] | 3121 Videos | NA | T | Depth | NA | 3D Skeleton Gaze ADOS scores |
| 3d-AD [48] | 100 Videos | 512 × 424 | T | Depth | U | Behaviors |