Scene segmentation algorithms intrinsically have got
To find an object's boundary if it is there or not.
Objective algorithms then extract quantitative features;
Constructing training sets is fine, but can you trust the teachers?

Comments