[Annotate] QUESTION - How does Labelbox enable annotation of semi-structured data?

Does Labelbox offer native support for annotating semi-structured data (i.e. CSV data)?

Screenshot attached above.

Use case would be to bring in a CSV, and then Annotate using a rules-based system (i.e. if Temperature > 90 Degrees F AND Humidity > 90%), then label/categorize the data row as ‘ASSET AT RISK OF FAILURE’.

Hey @Kush,

So at the moment, the only workaround I would have is to use tabulate to import table (will be recognised as text in Labelbox).

import labelbox as lb
import os
import tabulate as tb

api_key = None
client = lb.Client(api_key)

table_sample = [["Timestamp","Temperature","Humidity", "Wind Speed"],
            ["2023-09-01 00:00:00",90.5,90.7, 5 ],
            ["2023-09-01 01:00:00",20.2,36, 2.1],
            ["2023-09-01 02:00:00",21.7,20, 3.4],
            ["2023-09-01 03:00:00",24.9,39, 7.3],
            ["2023-09-01 04:00:00",19.0,40, 5]]

table_tb = tb.tabulate(table_sample)
print(table_tb)

dataset = client.create_dataset(name="test table", iam_integration=None)
create_data_row = dataset.create_data_row(row_data=table_tb)

If you have a CSV, you can workaround as well.

Filtering would have to be done before uploading.

It would look like :

Many thanks,
PT
Labelbox Support

2 Likes

I was just about to say work with the data before bringing it to labelbox - such as filter or conditional formatting - so that you don’t have to label tabular data. Basically be mindful of your workflow and order of operations. PT beat me to it though. Only by like 7 months, though.

2 Likes