You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I try to mask the origin image to train the moving object only.
At the same time, I set the background to black,which is [0,0,0]. In this way, when calculating the loss, the RGB loss will be 0.
However, the actual training result is more like a grayscale image (the red light has the color, but the yellow car body does not have color). As shown below, left is origin image, right is training eval image
So I would appreciate if there is an elegant way to train only dynamic vehicles or you can give me some advice.
Thanks for your great work!
I noticed that NeuRAD supports rendering only static background in
neurad-studio/nerfstudio/scripts/render.py
Line 738 in b7012ac
The text was updated successfully, but these errors were encountered: