Automatically build a face recognition training dataset 

How can I automatically build training datasets from my media footage? 

Media archives and broadcasters that have a lot of material with lower-thirds who would like to automate their data management

The Problem

Although training AIs is already very easy, there are so many unknown people in many archives that even uploading three images of the respective person leads to a massive time investment. Thousands of hours of video material thus develop into a data silo whose already contained information cannot be used. Is there a way with DeepVA to automatically create training data of specific people?

The Solution

Thanks to DeepVA’s Deep Collector, training data for face recognition can be obtained automatically. The AI of the Deep Collector recognizes lower-third titles, it checks the quality of the face present in the image and, if the quality is good, stores individual frames as training data together with the name extracted from the lower-third.

0 %

faster data acquisition

0 %


0 %

faster labelling

What results can be obtained?

Automated creation of a training database out of available video data

Permanent Extension

Integrated Quality Checks for best possible training data

Automatically build a face recognition training dataset 

Function overview

Face Dataset Creation

Using images of the faces and text information (e.g. from belly bands), Face Dataset Creation automatically creates entire datasets, without any manual labor.


Contact us

Do you have any questions?

Enter your email below and we will get back to you as soon as possible.

Related use cases

Take a look at our other use cases

Automatically build a speaker dataset 

How can I build up a dataset for my speaker recognition? 

Read More