nilportugues/shoe_searching_comparison

Coding practice: store/query data in MongoDB, processing .mat files, interactive plot results

Goal

An code example to search the shoe and plot two kind of shoes in the dataset UT Zappos50K .

Up to now, I only use the shoes in zappos-labels-fg.mat, because it contains comparative information about the similar pairs and reasons (Mturk).

Input: a dict of the 'key: value', i.e. {key1: value1, key2: value2, ...}

Output: An image showing two kind of shoes, one will match the conditions we speicify in the input (randomly from the mathcing result), and the other one is the pair of the matching one. The image will also show the attributes of the shoes and the comparison result.

The successful testing environment

System

• Ubuntu 16.04
• Python 2.7.14
• Install all the packages as list in the .py file in the src folder

MongoDB

We use MongoDB to store the data. So you need to install the MongoDB. And we use pymongo. Remember to start MongoDB service in a separate window.

mongo

Matlab2017b (Optional)

To store the dataset in the MongoDB, I need to process the .mat. Most of the .mat file in the dataset can be directly processed using spicy.io except the rationale-fg.mat. This file is a little like h5df. So I process it in Matlab and convert the information to ration.txt before processing in python.

Notes

I change the folder name of 'ut-zap50k-images-square\Shoes\Oxfords\Neil M%2E' to 'ut-zap50k-images-square\Shoes\Oxfords\Neil M.' for some reasons.