← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9003 · 生成模型:GPT-6

典型视角造成摄影偏差

理解等级:中级

人偏好拍摄容易辨认的典型视角,这使照片中的物体姿态分布区别于实际遭遇的姿态。书中十二物体、十二视角实验关联了偏好与识别速度准确率。构图、地平线、遮挡、天气和时间也影响照片分布;设计数据集要明确目标任务会遇到哪些非典型视角。

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

      In a beautiful study, Steve Palmer et al [371] showed participants photographs of objects with different orientations in order to measure what orientations they preferred. For each object there were 12 different orientations. Participants had consistent preferrences for particular viewpoints for each object. Interestingly, those preferences also correlated with how easy it was to recognize each object. In another experiment from the same study, participants

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第35章《数据偏差与分布漂移》(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/c035b.xhtml) - 结论依据:摘录中关于典型视角造成摄影偏差的定义、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:In a beautiful study, Steve Palmer et al [371] showed participants photographs of objects with different orientations in order to measure what orientations they preferred. For each object there were 12 different orientations. Participants had consistent preferrences for particular viewpoints for each object. Interestingly, those preferences also correlated with how easy it was to recognize each object. In another experiment from the same study, participants