知识卡片

普通读书笔记卡 · 8977 · 生成模型:GPT-6

颜色分组不保证空间连通

理解等级:中级

按颜色聚类再把中心颜色画回图像,可以显示量化区域,但同一簇可散布在多个不相连位置。簇编号不是天然的对象实例编号,也不保证语义一致。需要连续区域或实例时,还须加入空间、边界或其他结构约束。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      K-means tries to find clusters in this scatter plot (and, if you prefer a probabilistic interpretation, it can be thought of as fitting a mixture of Gaussians to the datapoints, which aims to approximate the distribution from which these pixels were sampled). We can segment via k-means by simply assigning each pixel to the cluster assignment of its color value. Below we color each pixel

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第31章《知觉分组》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)/_epub-src/OPS/xhtml/c031.xhtml) - 结论依据:摘录中关于颜色分组不保证空间连通的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。 - 原始内容:K-means tries to find clusters in this scatter plot (and, if you prefer a probabilistic interpretation, it can be thought of as fitting a mixture of Gaussians to the datapoints, which aims to approximate the distribution from which these pixels were sampled). We can segment via k-means by simply assigning each pixel to the cluster assignment of its color value. Below we color each pixel