知识卡片

普通读书笔记卡 · 9047.b · 生成模型:GPT-6

通用编码解码不自带几何保证

理解等级:中级

编码器提取上下文、解码器恢复分辨率,跳连传递局部细节,是单图深度的一种通用架构。输出形状像深度图不保证正值、平面性或相机一致性;这些性质由参数化、损失、数据与附加约束共同决定。

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

      The function, hθ, is usually implemented with a neural network as shown in figure 43.4. The architecture illustrated in the figure is a encoder-decoder with skip connections. The input is an red-green-blue (RGB) image and the output is a depth map. The architecture shown is very generic and there is nothing specific to the fact that it will be trained to uniquely estimate depth, but

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第43章《单图深度估计》(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/c043.xhtml) - 结论依据:摘录中关于通用编码解码不自带几何保证的定义、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:The function, hθ, is usually implemented with a neural network as shown in figure 43.4. The architecture illustrated in the figure is a encoder-decoder with skip connections. The input is an red-green-blue (RGB) image and the output is a depth map. The architecture shown is very generic and there is nothing specific to the fact that it will be trained to uniquely estimate depth, but