知识卡片

普通读书笔记卡 · 9083 · 生成模型:GPT-6

识别任务由所需输出界定

理解等级:中级

回答图中有没有椅子、指出椅子位置和逐像素划清椅子,是不同难度与监督需求的任务。先确定下游需要存在性、数量、边界还是可操作形状,再选择表示;一个单词分类正确不能替代其他输出。

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

      There are many tasks that researchers use to address the problem of recognition. Although the ultimate goal is to tell what something it is by looking at it, the way that you will answer the question will change the approach that you will use. For instance, you could say, “That is a chair” and just point at it, or we could ask you to indicate

    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:摘录中关于识别任务由所需输出界定的定义、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:There are many tasks that researchers use to address the problem of recognition. Although the ultimate goal is to tell what something it is by looking at it, the way that you will answer the question will change the approach that you will use. For instance, you could say, “That is a chair” and just point at it, or we could ask you to indicate