Skip to content

Fashion Mnist image classification using cross entropy and Triplet loss

Notifications You must be signed in to change notification settings

EmnamoR/FashionMnist

Repository files navigation

Fashion Mnist Image classification

For this experiment, I am going to use the Fashion MNIST dataset that consists of Zalando’s article images which is a set of 28x28 greyscale images of clothes, a drop-in replacement for the MNIST dataset. This is how the dataset looks like:
Mnist fashion data sample

Getting Started

These instructions will get you a copy of the project up and running on your local machine .

Prerequisites

  • python3.6 installed (app tested on3.6 only)
  • Have virtualenv installed on your machine
  • create virtual environment conda create --name adver or virtualenv adver --python python3
  • activate virtual env conda activate adver or source path_to_virtualenv/adver/bin/activate

Installing

There is 2 releases that one can install

  • First relase use crossEntropyLoss + Hyperparameter tuning
    git clone -b v1.0 https://github.com/EmnamoR/AdverFashionMnist.git
  • Second release uses Triplet Loss/ CrossEntropyLoss + Hyperparameter tuning
    git clone -b v1.2 https://github.com/EmnamoR/AdverFashionMnist.git

pip install -r requirements.txt

Run


To run this app

pythton train.py

Documentation

please find the full documentation of v1.1 release: here
please find the full documentation of v1.3 release(triplet loss release): here