Open InuyashaLee opened 5 days ago
注意到pose中没有使用全部的脸部关键点,只使用了眼睛耳朵鼻子。我尝试加上全部的脸部关键点,但生成的人脸部分变得更加奇怪。推测模型训练的时候pose就没有使用全部的脸部关键点,对吗?请问这样做的原因是什么呢?
Hi, we didn't train with the key points, because 1. Fair comparison with existing methods. 2. It is difficult to align the key points, when aligned the human posture.
注意到pose中没有使用全部的脸部关键点,只使用了眼睛耳朵鼻子。我尝试加上全部的脸部关键点,但生成的人脸部分变得更加奇怪。推测模型训练的时候pose就没有使用全部的脸部关键点,对吗?请问这样做的原因是什么呢?