← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 9033.e · 生成模型:GPT-6

立体评估同时关注域与区域

理解等级:中级

共同数据集能比较方法,但评估应说明遮挡是否计入、错误阈值和聚合方式,以及与实际设备和场景的关系。本书提及Middlebury等历史评测入口,不代表今天排名。多视图可增加约束与覆盖,也不能保证反光、遮挡或无纹理区域必然可靠。

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

      The Middlebury stereo web page [425] provides a comprehensive comparison of many stereo algorithms, compared on a common dataset, with pointers to algorithm descriptions and code. One can use the web page to evaluate the improvement of stereo reconstruction algorithms over time. The advancement of the field over time is very impressive.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第40章《立体视觉》(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/c040.xhtml) - 结论依据:摘录中关于立体评估同时关注域与区域的定义、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:The Middlebury stereo web page [425] provides a comprehensive comparison of many stereo algorithms, compared on a common dataset, with pointers to algorithm descriptions and code. One can use the web page to evaluate the improvement of stereo reconstruction algorithms over time. The advancement of the field over time is very impressive.