IoU IoU is Intersection over Union. Let's we have two overlappig boxes. $IoU = \frac{Intersection}{Union} = \frac{Intersection}{Area1 + Area2 - Intersction}$. When two bounding boxes perfectly match, IoU = 1. When two boxes are far away, IoU=0. IoU of two boxes is $\frac{0.25}{1.75}$. If IoU is 1, two boxes is same. So IoU means how similar the two boxes are. However IoU has a plateau making it ..

Motivation CNN has too many filters because CNN stack convolution layers. It means CNN has too many feature maps, parameters. It makes over fitting the network. So we need to decrease parameters. And CNN + FC has too many parameters, and make disappear information of position. Main Idea Average pooling Let's see average pooling. max average makes x([1,4,4,3]) to [1,2,2,3] with pool_size=2, strid..