thu-vis / OoDAnalyzer

A web-based tool for analyzing out-of-distribution samples.
42 stars 2 forks source link

OoDAnalyzer

Codes for the interactive analysis system, OoDAnalyzer, described in our paper "OoDAnalyzer: Interactive Analysis of Out-of-Distribution Samples" (TVCG 2021).

Online demo: http://visgroup.thss.tsinghua.edu.cn:8183/

Requirements

anytree==2.8.0
cffi==1.14.0
fastlapjv==1.0.0
Flask==1.1.2
matplotlib==3.1.3
numpy==1.18.4
Pillow==7.1.2
scikit-learn==0.22.1
scipy==1.4.1

Tested on Windows.

Usage Example

Step 1: create a folder data/ in the root folder.

Step 2: download demo data from Baiduyun(Link: here, password: 7nen) or Google Drive (Link: here, no password), and unpack it in the folder data/.

Step 3: setup the system:

python server.py

Step 4: visit http://localhost:8183/ with a browser.

Citation

If you use this code for your research, please consider citing:

@article{chen2021oodanalyzer,
  author={Chen, Changjian and Yuan, Jun and Lu, Yafeng and Liu, Yang and Su, Hang and Yuan, Songtao and Liu, Shixia},
  journal={IEEE Transactions on Visualization and Computer Graphics}, 
  title={{OoDAnalyzer}: Interactive Analysis of Out-of-Distribution Samples}, 
  year={2021},
  volume={27},
  number={7},
  pages={3335-3349}}

Contact

If you have any problem about our code, feel free to contact

or describe your problem in Issues.