知识卡片
视觉组件共享不等于外观相同
桌腿和椅腿可共享功能与结构角色,却不必具有完全相同的像素外观。组合表示要允许部件实例变化和关系差异,不能简单把视觉理解等同于固定词汇的字符串拼接。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 目标识别
原书出处 · 1 处
- · 查看原书依据
One important aspect of his theory is that objects are formed by compositing simpler elements (Geons), which are shared across many different object classes. Compositionality in the visual world is not as strong as the one found in language, where a fix set of words are composed to form sentences with different meanings. In the visual world, components shared across different object classes will have
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:摘录中关于视觉组件共享不等于外观相同的定义、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:One important aspect of his theory is that objects are formed by compositing simpler elements (Geons), which are shared across many different object classes. Compositionality in the visual world is not as strong as the one found in language, where a fix set of words are composed to form sentences with different meanings. In the visual world, components shared across different object classes will have