This is not a question about dlib, but a question on how to implement a feature on dlib.
Dlib has a virtual class dlib::tensor
, and an implementation of it named dlib::alias_tensor_instance
. Objects of this class are constructed by calling operator()(dlib::tensor &original, size_t offset)
on the factory class dlib::alias_tensor
, that will create a dlib::alias_tensor_instance
pointing to a internal slice of the original
tensor, that can be used wherever a tensor is required, without actually copying it.
Since it is very common to const dlib::tensor&
to be passed around, I am interested on a overload of the factory operator taking a const:
<something> operator()(const dlib::tensor &original, size_t offset);
Trouble is, neither I nor the author knows how it can be done properly. Do you guys knows of any solution? Here is the GitHub issue.
Aucun commentaire:
Enregistrer un commentaire