oliveredholm Goto Github PK

.tmux

Need to save this somewhere

a-algorithm

An implementation of the A* algorithm in Python.

alt-text-embeddings

Module for creating embeddings of the alt text associated with an image on the web, intention is to combine this with a Siamese CNN to learn good image -> text representations. The dataset is automatically created by using the method I implemented here: https://github.com/OliverEdholm/automatic-web-icon-dataset

automatic-click-heatmap-dom-voting

Trying to achieve general web element selectors. Experimented with a random forest-inspired approach to have 100's of CSS selectors to select an element, instead of one. Also, built an environment to swiftly qualitatively test various element selector algorithms on unseen websites through overlaying "click heatmaps" over web page screenshots, displaying the confidence scores of the element selector algorithm.

automatic-web-decoration-dataset

Scraping top N websites on the Alexa ranking, downloading all images, filtering based on if the associated alt text is provided, but empty. If an alt attribute is provided but it's empty it means that it's mostly a decoration and is non-relevant to get the actual context of the website.

automatic-web-icon-dataset

Scraping top N websites on the Alexa ranking, getting all images, filtering based on if the image is a natural photograph or a computer graphic, and on how big the image is. Including alt-text if it follows a set of criterion.

automatic-web-main-role-dataset

Automatically create dataset for extracting out main content of a website by looking at visual features. All of the scraping is happening at the Google Cloud Platform.

badmintonswedenapi

API for accessing Swedish Badminton related data easily.

convolutional-autoencoder

A convolutional autoencoder made in TFLearn.

django-heroku-template

Django template for developing good web apps that are being uploaded to Heroku for hosting.

dotfiles

fut

fut is a simple library for managing Fifa 18 Ultimate Team.

heroku-django-template

A Django 2.0 base template featuring all recommended best practices for deployment on Heroku and local development.

icon-alt-text-image-retrieval

Automatically providing alt text to graphical symbols by doing nearest neighbor search on simple shape image descriptors. Every nearest neighbor candidate has an alt text associated with it. The dataset is automatically created by using the method I implemented here: https://github.com/OliverEdholm/automatic-web-icon-dataset

icon-generative-adversarial-network

Code used for generating icons automatically through GAN, the initial implementation was taken from *carpedm*. Dataset used was generated by me through scraping the web + filtering images at: https://github.com/OliverEdholm/automatic-web-icon-dataset

image-retrieval

Image retrieval program made in Tensorflow supporting VGG16, VGG19, InceptionV3 and InceptionV4 pretrained networks and own trained Convolutional autoencoder.

ionic-native

Native features for mobile apps built with Cordova/PhoneGap and open web technologies. Complete with TypeScript support. The successor to ngCordova. Pairs exquisitely with a nice bottle of Ionic Framework.

jsdom

JSDOM, but much faster, modified for project doing analysis on large parts of the web

jupyterlab-vim

:neckbeard: Vim notebook cell bindings for JupyterLab

neural-number-sorter

Neural network that successfully sorts X-bit numbers that are Y numbers long.

neural-quote-generator

A lstm-char-rnn generator (neural network) made in TFLearn generating quotes.

nvda-addon

NVDA add-on that describes images

perceptron-network

An one-layer perceptron network made for learning.

python-algorithms

A repository of algorithms I've implemented in Python

python-bing-web-api

A Bing web API for python. It only works for v1 and you need to have an API key to use the API.

scholarly

Retrieve author and publication information from Google Scholar in a friendly, Pythonic way

stock-market-probabilities-deep-learning

Tries to predict if a stock will rise or fall with a certain percentage through giving probabilities of what events it thinks will happen.

super-resolution.benckmark

Benchmark and resources for single image super-resolution algorithms

tencent-ml-images

Largest multi-label image database; ResNet-101 model; 80.73% top-1 acc on ImageNet

tensorflow-easy-seq2seq

A tool that allows you to easily train a Seq2Seq model, get the embeddings and the outputs without having much knowledge about deep learning.