MATLAB code for "Dress like a Star: Retrieving Fashion Products from Videos" paper. This version of the code is not optimized to run efficiently.

• MATLAB
• OpenCV (3.3.0)
• mexopencv

1. Add mexopencv MEX functions to MATLAB path:

addpath('/path/to/mexopencv');
addpath('/path/to/mexopencv/opencv_contrib');

2. Training. Run training.m to extract features and create the model.

videoDir = '/path/to/videos';
dataDir  = '/path/to/data';
training(videoDir, dataDir);

3. Testing. Run query2frame.m to match a query image with a dataset frame.

query = imread('/path/to/query.png');
dataDir  = '/path/to/data';
query2frame(query, dataDir);

We have developed a small GUI demo in MATLAB for Unix systems. To play with it run the demo_gui.m code.

This code is the MATLAB's implementation of the paper "Dress lika a Star: Retrieving Fashion Products from Videos" published in the Proceedings of the Computer Vision for Fashion Workshop at ICCV 2017.

@InProceedings{GarciaICCVW2017,
author    = {Noa Garcia and George Vogiatzis},
title     = {Dress like a Star: Retrieving Fashion Products from Videos},
booktitle = {Proceedings of the International Conference on Computer Vision Workshops (ICCVW)},
year      = {2017},
}