← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 9033.c · 生成模型:GPT-6

三维聚合以算力换一致性

理解等级:中级

3D CNN在代价体上联合处理空间与视差邻域,承担传统方法正则化与匹配聚合的部分作用。分辨率和视差范围扩大都会增加时间、内存;迭代方法可用不同结构减少显式3D过滤开销。效果与成本应在相同评估条件下比较,不能只凭网络名称判断。

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

      Train and apply a 3D convolutional neural net (CNN) to aggregate (process) the costs over the 3D cost volume in order to estimate a single, best disparity estimate for each pixel position. This performs the function of the regularization step of nonneural-network approaches, such as SGM [207], but the performance is better for the neural network approaches.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第40章《立体视觉》(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/c040.xhtml) - 结论依据:摘录中关于三维聚合以算力换一致性的定义、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:Train and apply a 3D convolutional neural net (CNN) to aggregate (process) the costs over the 3D cost volume in order to estimate a single, best disparity estimate for each pixel position. This performs the function of the regularization step of nonneural-network approaches, such as SGM [207], but the performance is better for the neural network approaches.