“Human Cloud” used to create videos of everyday activities, from getting into cars to exiting rooms.

Tannah Cantrell
February 29, 2024

About the Company

Researchers at a US-based research institute were looking to help their AI develop quality algorithms and learn to convincingly label more human actions on a mass scale. They also needed to train long-range biometric identification.

As participants were needed on such a large scale, this company made the decision to reach out to Twine AI. The project went on to commission Twine’s network of freelancers, referred to as the “human cloud”, to create videos of everyday activities, from getting into cars to exiting buildings.

The Solution

From all of the data collected, researchers at this US-based research facility were able to create an open-source video library made available to the international research community. Before this project, YouTube was the primary source of information for this type of research. Subsequently, this means that video data is often low quality, or not fit for purpose video.

This project, in collaboration with Twine, has rivaled YouTube. Businesses can now use this extensive, open-source video library, to develop quality algorithms and replicate more convincing human movements.


Not only has this project reached milestones for video datasets, but it also provided an estimated 20,000 freelancers with the opportunity to participate in $1,000,000 worth of work, during the COVID-19 outbreak. With 86% of freelancers working from home at the time, micro-projects like this continue to help supplement a freelancer's income.

In addition to properly managing the project and recruiting freelancers to generate the dataset, Twine's secure vault payment system was used to manage all payments to the freelancers internationally.

Related Customer Stories