How to train your models with IO Annotator

...
Sascha Heyer
Machine Learning

IO Annotator builds a platform around people and data to solve the most significant challenges in machine learning. Data is key to successful AI solutions we’ve built a modern Annotation Platform that helps you to do that better.

Our clients are by nature from very different sectors with even more variation in their machine learning approaches. Therefore we provide you three different ways to integrate our platform into your machine learning environments.

Exports

We start with the most basic way, a plain export of your annotated training data. If you are a non-technical user this is the best approach. Go to your dataset dashboard, navigate to export and click download. We process your dataset and notify you when the export is ready. You can then simply download your annotated data.

Depending on the type of annotation we offer various types of export formats like CSV and JSON.

Webhooks

Webhooks are a great way if you want to receive realtime updates of your annotated data. Instead of polling, our API for data our webhooks sends events each time an action is performed. We will send a POST request to your defined endpoint. This request contains all the required information.

Webhooks can be enabled on datasets.

API

The API provides you a way to get the data in a more technical manner this enables you to integrate our platform into your machine learning environments.

The APIKey and your DatasetID are just examples, replace those with your values and send the request like that:

URL = "https://api.ioannotator.com/api/export"
# parameter
apikey = "PZK888A-AEQMDPY-MKKPU2Q-JZ786NX"
dataset = "5758665286207488"
# defining a params dict for the parameters to be sent to the IO Annotator API
PARAMS = {'apikey':apikey, 'dataset': dataset}
# sending get request and saving the response (annotations) as response object
r = requests.get(url = URL, params = PARAMS)
# extracting data in json format
data = r.json()

PythonSDK

The IO Annotator team is currently working on a PythonSDK to offer you an even more comfortable way to work with data.