kl-loss

This is only the caffe implementation for the KL loss in the paper:

Bounding Box Regression with Uncertainty for Accurate Object Detection

@inproceedings{klloss,
title={Bounding Box Regression with Uncertainty for Accurate Object Detection},
author={He, Yihui and Zhu, Chenchen and Wang, Jianren and Savvides, Marios and Zhang, Xiangyu },
booktitle={2019 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
year={2019},
organization={IEEE}
}

How to write the prototxt?

if you have three inputs (not include weights)

layer {
  name: "loss_kl"
  type: "KlLoss"
  bottom: "coor"
  bottom: "theta"
  bottom: "label"
  top: "kl_loss"
  propagate_down: 1
  propagate_down: 1
  propagate_down: 0
  loss_weight: 1
}

if you have four inputs (include weights)

layer {
  name: "loss_kl"
  type: "KlLoss"
  bottom: "coor"
  bottom: "theta"
  bottom: "label"
  bottom: "weight"
  top: "kl_loss"
  propagate_down: true
  propagate_down: true
  propagate_down: false
  propagate_down: false
  include {
    phase: TRAIN
  }
}