SyRIP
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/ostadabbas/infant-pose-estimation
下载链接
链接失效反馈官方服务：
资源简介：
该数据集名为SyRIP，包含700张真实婴儿图像和1000张合成婴儿图像，同时还通过将SMIL模型拟合到真实图像生成了额外的变体。在600个弱地面真实3D标签上进行训练，并在100张真实图像上进行测试。整个数据集规模为1700张图像（其中700张为真实图像，1000张为合成图像），其任务是进行3D姿态估计。
This dataset, named SyRIP, contains 700 real infant images and 1000 synthetic infant images. Additionally, extra variants were generated by fitting the SMIL model to the real infant images. The dataset is trained on 600 weakly supervised 3D ground-truth labels and tested on 100 real infant images. In total, the dataset consists of 1700 images (700 real and 1000 synthetic), and its target task is 3D pose estimation.
