pytorch-retinanet's Introduction

RetinaNet-pytorch

This repo aims to implement a "Focal loss for Dense Object Detection"

Requirements

Training

# Select the script that you want to train for reproducing a results
./retina_ce_sgd_0.001.sh
# For the focal loss use ./retina_focal_sgd_0.0001.sh

You can see the details in trainer.py

VOC Dataset

#####Download VOC2007 trainval & test

# specify a directory for dataset to be downloaded into, else default is ~/data/
sh data/scripts/VOC2007.sh # <directory>

#####Download VOC2012 trainval

# specify a directory for dataset to be downloaded into, else default is ~/data/
sh data/scripts/VOC2012.sh # <directory>

TODO

Building RetinaNet
Apply cross entropy loss and focal loss(still checking it's works or not)
Comparing btw CE loss and focal loss
Report results on VOC
- currently i get 50mAP on VOC2007. There is some problem that i don't know yet.
Report results on COCO
Use relative path for easy reproducing of result

References

Keras version
- Easy to read and very active repository. This repo drive me complete my code
Another pytorch version
- Almost all code are followed from this repository, but i am not sure this repo works

Author

jtlee at omnious dot com

Feel free to contact me about code.

pytorch-retinanet's People

Contributors

Stargazers

Watchers

pytorch-retinanet's Issues

Not work on VOC dataset

I experiment with my own dataset which produce very good results like SSD_v2.
But currently not working on VOC dataset.
I think data generation or decoding part have some problem.

i will check out very soon and close this issue.
Anyone help to fix it will thank you.

Recommend Projects