Tensorflow Implementation of the paper Chen et al. Semi-Supervised Deep Learning with Memory, ECCV2018.
- Python 2.7.
- Tensorflow version >= 1.4.0.
- Download and prepare datasets:
bash scripts/download_prepare_datasets.sh
- Convert image data to tfrecords:
bash scripts/convert_images_to_tfrecords.sh
For example, to train and test on svhn
, run the following command.
bash scripts/train_svhn_semi.sh
Please refer to the following if this repository is useful for your research.
@inproceedings{chen2018semi,
title={Semi-Supervised Deep Learning with Memory},
author={Chen, Yanbei and Zhu, Xiatian and Gong, Shaogang},
booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
year={2018}
}
This project is licensed under the MIT License - see the LICENSE.md file for details.