知识卡片
条件生成保留未指定变化
文本通常只约束场景部分属性,噪声变量提供其他细节的变化,使同一描述可对应多张图。生成结果应检查是否满足已指定条件与基本结构,而非要求与唯一参考图逐像素相同;随机性本身不证明所有样本都合理。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 视觉与语言
原书出处 · 1 处
- · 查看原书依据
One amazing ability of modern generative models is to synthesize an image just from a text description, a problem called text-to-image. This can be achieved using any of the conditional generative models we saw in chapter 34. We simply have to find an effective way to use text as the conditioning information to these models. Figure 51.7 shows how this problem looks as a mapping
参考来源
- 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第51章《视觉与语言》(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/c051.xhtml)
- 结论依据:摘录中关于条件生成保留未指定变化的定义、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:One amazing ability of modern generative models is to synthesize an image just from a text description, a problem called text-to-image. This can be achieved using any of the conditional generative models we saw in chapter 34. We simply have to find an effective way to use text as the conditioning information to these models. Figure 51.7 shows how this problem looks as a mapping