Skip to content

关于训练数据的处理问题 #8

Open

Description

@raoqiang

论文中说, 训练数据分别取仰角, -20, 0, 20, random, 偏角360度均匀划分为24份, 这样会不会存在采样不均匀的情况? 仰角大部分都设置到 -20, 0, 20 这个范围的, 这个是出于什么考虑呢?

Activity

  1. ZebinHe commented on Aug 19, 2025

    @ZebinHe
    Owner

    Hi~因为大部分情况下,用户输入的参考图像(或者文生图模型生成的参考图像)都是偏正面的视图,可能会带有轻微的俯仰,少数情况下才出现大角度俯仰角,出于这种情况设置的训练数据相机参数

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions