bbbc006.py 文件源码

python
阅读 34 收藏 0 点赞 0 评论 0

项目:dcan-tensorflow 作者: lisjin 项目源码 文件源码
def get_dice_coef(logits, labels):
    """Compute dice coefficient.
    Args:
        logits: Softmax probability applied to fuse layers.
        labels: Correct annotations (0 or 1).
    Returns:
        Mean dice coefficient over full tensor.

    Source:
        https://github.com/zsdonghao/tensorlayer/blob/master/tensorlayer/cost.py#L125
    """
    smooth = 1e-5
    inter = tf.reduce_sum(tf.multiply(logits, labels))
    l = tf.reduce_sum(logits)
    r = tf.reduce_sum(labels)
    return tf.reduce_mean((2.0 * inter + smooth) / (l + r + smooth))
评论列表
文章目录


问题


面经


文章

微信
公众号

扫码关注公众号