jirenz / HiDDeN

HiDDeN: Hiding Data with Deep Networks.

HiDDeN: Hiding Data with Deep Networks.

HiDDeN: Hiding Data with Deep Networks. arxiv ECCV 2018
Jiren Zhu*, Russell Kaplan*, Justin Johnson, Li Fei-Fei
*: These authors contributed equally

Warning: This repo is still WIP, let us know if you encounter bugs or issues.

Data

A sample dataset is provided in data/yuv_coco_debug.t7. Afte you download train2014 of COCO dataset, you can use HiDDeN/coco_prep.lua to generate the training dataset.

Running models

• With the testing dataset. You can run

th main.lua --develop --name test-run --type float

, training results will be written to checkpoints/test-run.

• If you have cuda setup. You can run

th main.lua --develop --name <experiment name>

instead, as --type cuda is default.

If you have the full dataset, remove the develop flag.

Pretrained models

Coming soon...

Other implementations

• Pytorch implementation by Ando Khachatryan: https://github.com/ando-khachatryan/HiDDeN.