https://github.com/csailvision/ibd
IBD: Interpretable Basis Decomposition for Visual Explanation
https://github.com/csailvision/ibd
Last synced: 10 months ago
JSON representation
IBD: Interpretable Basis Decomposition for Visual Explanation
- Host: GitHub
- URL: https://github.com/csailvision/ibd
- Owner: CSAILVision
- Created: 2018-07-11T12:26:30.000Z (almost 8 years ago)
- Default Branch: master
- Last Pushed: 2018-11-28T15:17:53.000Z (over 7 years ago)
- Last Synced: 2025-04-17T02:02:45.405Z (about 1 year ago)
- Language: Python
- Size: 5.69 MB
- Stars: 52
- Watchers: 6
- Forks: 14
- Open Issues: 3
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# IBD: Interpretable Basis Decomposition for Visual Explanation
## Introduction
This repository contains the demo code for the ECCV'18 paper "Interpretable Basis Decomposition for Visual Explanation".
## Download
* Clone the code of Network Dissection Lite from github
```
git clone https://github.com/CSAILVision/IBD
cd IBD
```
* Download the Broden dataset (~1GB space) and the example pretrained model. If you already download this, you can create a symbolic link to your original dataset.
```
./script/dlbroden.sh
./script/dlzoo.sh
```
Note that AlexNet models work with 227x227 image input, while VGG, ResNet, GoogLeNet works with 224x224 image input.
## Requirements
* Python Environments
```
pip3 install numpy sklearn scipy scikit-image matplotlib easydict torch torchvision
```
Note: The repo was written by pytorch-0.3.1. ([PyTorch](http://pytorch.org/), [Torchvision](https://github.com/pytorch/vision))
## Run IBD in PyTorch
* You can configure `settings.py` to load your own model, or change the default parameters.
* Run IBD
```
python3 test.py
```
## IBD Result
* At the end of the dissection script, a HTML-formatted report will be generated inside `result` folder that summarizes the interpretable units of the tested network.
## Train Concept Basis
* If you want to train the concept basis, delete the pretrained files first.
```
rm result/pytorch_resnet18_places365/snapshot/14.pth
rm result/pytorch_resnet18_places365/decompose.npy
```
* Run the train script.
```
python3 train.py
```
* Then run IBD.
```
python3 test.py
```
## Reference
If you find the codes useful, please cite this paper
```
@inproceedings{IBD2018,
title={Interpretable Basis Decomposition for Visual Explanation},
author={Zhou, Bolei* and Sun, Yiyou* and Bau, David* and Torralba, Antonio},
booktitle={European Conference on Computer Vision},
year={2018}
}
```