This page is only partially interactive. Since this is a static HTML page, only front-end interactivity works. This means you can click buttons, but the relevant python-level responses to those actions won’t occur.

Classifying images, text, and arbitrary input

Classification is a standard task in data labelling, and ipyannotations has support for varying input types.

All classification widgets share some parameters and arguments, such as:

  • options: the classes you’re assigning

  • allow_freetext: whether to allow free text entry of new class labels

  • max_buttons: depending on the labelling job, it may get too unwieldy to have every option as a button. Setting this allows the widget to switch to a different method of class selection.

All classification widgets also respond to hotkeys: you can use the numeric keys 1 – 0 to select an option, with keys mapped to the classes in order. You can also use “Backspace” to undo an option, and in the case of multi-class widgets, you can use “Enter” to submit.

Image classification

This is a standard task in machine learning, and the data required for these tasks can be generated quickly with ipyannotations.

import ipyannotations.images

widget = ipyannotations.images.ClassLabeller(
    options=['baboon', 'orangutan'], allow_freetext=True)

Text classification

The text classification widget works much the same way. Imagine you want to classify things as spam or not spam:

import ipyannotations.text

widget = ipyannotations.text.ClassLabeller(
    options=['spam', 'not spam'], allow_freetext=False)
    "Greetings! Your esteemed research would be suitable "
    "for publication in our scientific journal.")

There is also a special classification widget for sentiment in the text module:

import ipyannotations.text

widget = ipyannotations.text.SentimentLabeller()
widget.display("You look nice today.")

Multiple class labels (multiclass)

You can also assign multiple classes to a data point, with the MulticlassLabeller. It exists for both images and text, and allows the user to toggle multiple classes, then click submit (or hit Enter):

import ipyannotations.images

widget = ipyannotations.images.MulticlassLabeller(
    options=['baboon', 'mammal', 'toucan', 'bird'], allow_freetext=True)

Arbitrary data with self-written display functions

In addition to the image and text widgets, you can build a custom classification widget by using the widgets from the ipyannotations.generic submodule. (In fact, the image and text widgets are just wrappers around this).

The display function for these widgets can be anything that displays output in a jupyter notebook.

For example, if you wanted to classify graphs of points into periodic and non-periodic:

import ipyannotations.generic
import numpy as np
import matplotlib.pyplot as plt

plt.ioff() # turn off default inline plotting

def plotting_function(data):
    fig, ax = plt.subplots(1, 1)

widget = ipyannotations.generic.ClassLabeller(
    options=['periodic', 'non-periodic'], allow_freetext=False,