README.md

RetinaNet

This is a tensorflow implementation for the paper Focal Loss for Dense Object Detection by FACEBOOK AI Research which is shortened as RetinaNet

References:

• Focal Loss for Dense Object Detection
• Feature Pyramid Networks for Object Detection

TO DO :

Requirements:

keras
tensorflow>=1.4.1
bunch

Usage :

  chmod 777 get_dependencies.sh
  ./get_dependencies.sh
  chmod 777 run.sh
  ./run.sh

Results:

Results on VOC :

Results on COCO :

Results on Kitti :

Other Implementations :

Pytorch

• PyTorch-FPN

Tensorflow

• FastFPN
• R-DFPN: Rotation Dense Feature Pyramid Networks
• Feature Pyramid Networks for Object Detection

Mxnet

• focal-loss
• Retina-Net
• Feature Pyramid Networks for Object Detection on mxnet

Caffe

• Reproduction of Focal-loss on Caffe
• loss layer implementation
• Caffe implementation of "Focal Loss" for SSD.
• Feature Pyramid Network on caffe