microsoft/View-Adaptive-Neural-Networks-for-Skeleton-based-Human-Action-Recognition

What is the reason to subtract the pixel values by 110?

sksenthilkumar opened this issue · 1 comments

In the preprocessing step, all the pixel values of the pseudo image have been subtracted by 110.
If we remove this function will it affect the performance?

It is similar to the normalization with mean. In fact, you can ignore this step because of the little impact on the final result.