Skip to content

Latest commit

 

History

History
101 lines (70 loc) · 2.91 KB

README.md

File metadata and controls

101 lines (70 loc) · 2.91 KB
Raw

Outfit Compatibility Prediction and Diagnosis with Multi-Layered Comparison Network

Paper: Outfit Compatibility Prediction and Diagnosis with Multi-Layered Comparison Network

by Xin Wang, Bo Wu, Yueqi Zhong. Published at ACM MM 2019 in Nice, France.

[Paper] [Arxiv] [Demo]

Contents of this repository

  • mcn: Main program source code
  • data: Polyvore-T datasets based on Polyvore.
  • baselines: Compared baselines in our experiment
  • exp: Experiment details, scripts and results etc.

Requirements

Ubuntu 16.04, NVIDIA GTX 1080Ti (for batch size 16), python >= 3.5.2

torch==1.0.1
torchvision==0.2.1
networkx==2.4
opencv-python==4.2.0.32
matplotlib==2.2.2
scikit-learn==0.21.2

Usage

  1. Download the original Polyvore dataset, then unzip the file and put the image directory into data folders (or you can create a soft link for it).

  2. Train

    cd mcn
python train.py

  3. Evaluate

    python evaluate.py

  4. Visualize outfit diagnosis

    cd exp
python diagnosis.py

  5. Automatically revise outfit

    python revision.py

Prediction Performance

Pretrained model weights can be found in the links. The train, validation and test split is provided in data.

AUC FITB
Pooling 88.35 57.28
Concatenation 83.40 52.91
Self-attention 79.65 48.60
BiLSTM 74.82 46.02
CSN 84.90 57.06
Ours 91.90 64.35

Demo

A demo application is in the app directioy. You can run it locally by go to app directory then use command python main.py.

More guide can be found in here.

Cite

Please cite our paper if you use or refer this code:

@inproceedings{wang2019diagnosis,
  title={Outfit Compatibility Prediction and Diagnosis with Multi-Layered Comparison Network},
  author={Xin Wang, Bo Wu and Yueqi Zhong},
  booktitle={ACM International Conference on Multimedia},
  year={2019}
}