c++ - Does it matter what types of negative images you use for Haar-like feature trainning? -
i'm trying object tracking using opencv , read bit haar-like features. says need 5000 - 10000 negative images , more images use more accurate results be. wondering matter type of images use it?
for example, tracking objects inside room better negative images of indoor environments? or not matter because it's negatives?
i apologies if sounds silly question, want 100% sure make sure results can accurate possible.
thanks
from read doesn't matter if these negative images outdoor or indoor, think should have both of these types.
btw in link naotoshi seo says used 3000 negative samples, if planning on having 5000 10000 of them, can use both indoor , outdoor negatives. make sure there no objects of interest in these negative samples.
hope helps.
Comments
Post a Comment