Darknet: v2 anchors vs v3 anchors

Created on 15 Aug 2018  Â·  3Comments  Â·  Source: pjreddie/darknet

dumb question
i mean why do they look so different (ones in in the hundred yet the others are very small )?

Most helpful comment

v2 - for size of the last layer (size of final feature map)
v3 - for size of the first layer (size of neural network)

All 3 comments

yeah,i also finde this problem,does anyone can make it sense?

v2 - for size of the last layer (size of final feature map)
v3 - for size of the first layer (size of neural network)

ah , thank you @AlexeyAB

Was this page helpful?
0 / 5 - 0 ratings