Why the length of output bbox is 8 ?

[jay2002]

Thanks for this great work first.
I have question about the network structure, this is, why the length of output bbox is 8? If we use class-agonistic bbox regression, the length of bbox vector should be 4. Are the first 4 values regressed for background, and the last 4 values regressed for objects?

[YuwenXiong]

Yes, in agonistic bbox regression it is actually a 2-class bbox regression, the first 4 dims are for background and the last 4 dims are for foreground. In fact the first 4 dims will never be used.