知识卡片

普通读书笔记卡 · 9091.b · 生成模型:GPT-6

语义分割输出逐像素类别

理解等级:中级

语义分割为每个像素给出类别分数,常用编码器—解码器或共享局部分类实现。式50.20把输入写成x[n,m]过简,实际判断可依赖整图或邻域上下文,并非仅用该像素RGB识别对象。

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

      An object is something localized in space. There are other things that are not localized such as fog, light, and so on. Not everything is well-described by a bounding box (stuff, wiry objects, etc.). Instead we can try to classify each pixel in an image with an object class. Per-pixel classification of object labels, illustrated in figure 50.20, is referred to as semantic segmentation.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第50章《目标识别》(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/c050b.xhtml) - 结论依据:摘录中关于语义分割输出逐像素类别的定义、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:An object is something localized in space. There are other things that are not localized such as fog, light, and so on. Not everything is well-described by a bounding box (stuff, wiry objects, etc.). Instead we can try to classify each pixel in an image with an object class. Per-pixel classification of object labels, illustrated in figure 50.20, is referred to as semantic segmentation.