yukitsuji / 3D_CNN_tensorflow

KITTI data processing and 3D CNN for Vehicle Detection

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Bounding Box prediction

OneManArmy93 opened this issue · comments

Can anyone tell me how the bounding boxes are predcition with the model? and which function does that?
Thank you.

This architecture will provide you with eight points in the output, you can join the points accordingly to get the bounding box. In some cases neural network will provide with multiple bounding boxes, in that case you have to perform non maximun suppression.

@ansabsheikh9 can be kind enough to suggest how to join the points at the training process in order to predict the bbox? thank you

While training you dont need to provide the bounding box shape. You have to provide with regression targets for each instance of vehicle, like eight corners with centered zero.