Focal loss for Dense Object Detection
Clone or download
Latest commit d74e54d Feb 1, 2018
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
data/cache retina-net Sep 6, 2017
lib re init retinanet params Jan 13, 2018
model/pretrained_model retina-net Sep 6, 2017
output retina-net Sep 6, 2017
retinanet Update train_end2end.py Feb 1, 2018
vis fix bug Jan 12, 2018
vis_restore fix bug Jan 12, 2018
README.md Update README.md Sep 6, 2017
init.sh retina-net Sep 6, 2017
pascal_voc.yaml Update pascal_voc.yaml Feb 1, 2018
test.py fix bug Jan 12, 2018
train.py retina-net Sep 6, 2017

README.md

Retina-Net

Focal loss for Dense Object Detection

The code is unofficial version for RetinaNet in focal loss for Dense Object Detection. https://arxiv.org/abs/1708.02002

You can use the focal loss in https://github.com/unsky/focal-loss

usage

install mxnet v0.9.5

  1. download the dataset in data/

  2. download the params in https://onedrive.live.com/?authkey=%21AI3oSHAoAIbxAB8&cid=F371D9563727B96F&id=F371D9563727B96F%21102802&parId=F371D9563727B96F%21102787&action=locate

./init.sh

train & test

python train.py --cfg kitti.yaml

python test.py --cfg kitti.yaml