About the pretrained model for monkey hand keypoints detection

[Haixin-Gong]

Hello,

I am trying to detect the keypoints of monkey hands, and the code worked pretty well for the given example project which aims to achieve the same goals. That is inspring.

But now I have the recordings from only 3 cameras (which is different from the example project of 12 cameras). So my questions is if the pretrained model can be utilized to work for less than 12 cameras? (For example, I can flip or rotate the videos to create fake perspectives up to 12).

And I am wondering if the difference in camera setup wiill affect the estimation results?

Thanks.