知识卡片

普通读书笔记卡 · 9079.a · 生成模型:GPT-6

无监督光流仍需训练数据

理解等级:中级

本章把不需要光流真值的训练称为无监督,但网络仍需大量视频或图像对。应区分没有人工运动标签与完全没有训练数据;后者不符合本章后续对视频集合优化参数的定义。

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

      Learning-based approaches rely on many of the concepts we introduced in the previous chapters. We will differentiate between two big families of models: supervised models that learn to estimate motion using a database of training examples, and unsupervised models that learn to estimate motion without training data.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第49章《学习运动估计》(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/c049.xhtml) - 结论依据:摘录中关于无监督光流仍需训练数据的定义、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:Learning-based approaches rely on many of the concepts we introduced in the previous chapters. We will differentiate between two big families of models: supervised models that learn to estimate motion using a database of training examples, and unsupervised models that learn to estimate motion without training data.