SYSTEM AND METHOD FOR AUTOMATIC LANDMARK LABELING WITH MINIMAL SUPERVISION
    1.
    发明申请
    SYSTEM AND METHOD FOR AUTOMATIC LANDMARK LABELING WITH MINIMAL SUPERVISION 有权
    自动地标标签系统与方法与最小监控

    公开(公告)号:US20130243309A1

    公开(公告)日:2013-09-19

    申请号:US13892102

    申请日:2013-05-10

    IPC分类号: G06K9/66

    摘要: A system and method for estimating a set of landmarks for a large image ensemble employs only a small number of manually labeled images from the ensemble and avoids labor-intensive and error-prone object detection, tracking and alignment learning task limitations associated with manual image labeling techniques. A semi-supervised least squares congealing approach is employed to minimize an objective function defined on both labeled and unlabeled images. A shape model is learned on-line to constrain the landmark configuration. A partitioning strategy allows coarse-to-fine landmark estimation.

    摘要翻译: 用于估计大图像集合的一组地标的系统和方法仅使用来自集合的少量手动标记的图像,并且避免与手动图像标签相关联的劳动密集型和易出错的对象检测,跟踪和对准学习任务限制 技术 采用半监督的最小二乘法凝结方法来最小化在标记和未标记图像上定义的目标函数。 形状模型在线学习以约束地标配置。 分区策略允许粗略到精细的地标估计。

    System and method for automatic landmark labeling with minimal supervision
    2.
    发明授权
    System and method for automatic landmark labeling with minimal supervision 有权
    最小监督的自动地标标签系统和方法

    公开(公告)号:US08897550B2

    公开(公告)日:2014-11-25

    申请号:US13892102

    申请日:2013-05-10

    IPC分类号: G06K9/00 G06T7/00 G06K9/66

    摘要: A system and method for estimating a set of landmarks for a large image ensemble employs only a small number of manually labeled images from the ensemble and avoids labor-intensive and error-prone object detection, tracking and alignment learning task limitations associated with manual image labeling techniques. A semi-supervised least squares congealing approach is employed to minimize an objective function defined on both labeled and unlabeled images. A shape model is learned on-line to constrain the landmark configuration. A partitioning strategy allows coarse-to-fine landmark estimation.

    摘要翻译: 用于估计大图像集合的一组地标的系统和方法仅使用来自集合的少量手动标记的图像,并且避免与手动图像标签相关联的劳动密集型和易出错的对象检测,跟踪和对准学习任务限制 技术 采用半监督的最小二乘法凝结方法来最小化在标记和未标记图像上定义的目标函数。 形状模型在线学习以约束地标配置。 分区策略允许粗略到精细的地标估计。