This is the official PyTorch implementation of "What Images are More Memorable to Machines?" This repo currently supports the training and testing of the MachineMem predictor. Instructions for GANalyze are also provided. Codes for running the MachineMem measurer will be released after paper acceptance.
What Images are More Memorable to Machines?
Junlin Han, Huangying Zhan, Jie Hong, Pengfei Fang, Hongdong Li, Lars Petersson, Ian Reid
University of Adelaide and Data61-CSIRO and Australian National University
- Clone this repo:
git clone https://github.com/JunlinHan/MachineMem.git-
Install PyTorch 1.7 or above and other dependencies (e.g., torchvision, timm, numpy, scipy, pillow, scikit-learn).
-
You may build a new conda environment by running
conda env create -f environment.yml.
For predictor, please see ./predictor.
For GANalyze, please see ./GANalyze_instructions.
junlin.han@data61.csiro.au or junlinhcv@gmail.com
If you use our code or our results/findings, please cite our paper. Thanks!
@inproceedings{han2022machinemem,
title={What Images are More Memorable to Machines?},
author={Junlin Han and Huangying Zhan and Jie Hong and Pengfei Fang and Hongdong Li and Lars Petersson and Ian Reid},
booktitle={arXiv preprint arXiv:2211.07625},
year={2022}
}
Our code is developed based on official PyTorch ImageNet training code.