Home

Awesome

A PyTorch Implementation for Triplet Networks

This repository contains a PyTorch implementation for triplet networks.

The code provides two different ways to load triplets for the network. First, it contain a simple MNIST Loader that generates triplets from the MNIST class labels. Second, this repository provides a Triplet Loader that loads images from folders, provided a list of triplets.

Example usage:

$ python train.py

Tracking experiments with Visdom

This repository allows to track experiments with visdom. You can use the VisdomLinePlotter to plot training progress.

<img src="https://github.com/andreasveit/triplet-network-pytorch/blob/master/images/visdom.png?raw=true" width="400">

If this implementation is useful to you and your project, please also consider to cite or acknowledge this code repository.