Implementation of "Learning Deep Features for Discriminative Localization"
Download data with the script below. The script download tiny-imagenet data and transform to pickle and tfrecord. The pickle include datasets and meta information of label dictionary. The tfrecord contains datasets; train dataset, valid dataset, test dataset.
python3 download_data.py
- python3.4+
- pip3
- python packages - requirements.txt
- tensorflow 1.4+
$ jupyter lab
and open train.ipynb
or
use snippet_train_model.py
$ jupyter lab
and open load_and_test.ipynb