Data annotation, a booming sector!

February 12, 2020

Newsletter January 2020

Click here to see the newsletter in your browser

Data annotation: diving into the workings of AI

Behind the promises of artificial intelligence lies data. On the quality of the second one, the performance of the first one will depend. This is a major challenge that requires an ever-increasing amount of annotation work, in which the human and machine eyes work together. In recent years, many companies have entered this multi-billion dollar market. Focus on a booming sector that aims, in a way, to endow artificial intelligence with sight.

Until a few years ago, the work of annotation required little qualification: this is a man, this is a chair... But as the models are perfected, it becomes more and more complex. For example:

• When it used to be necessary to score only a few points on a face for a facial recognition algorithm, today it takes more than 200, including 8 for each eyebrow and 20 for the lips.

A demanding, time-consuming job (80% of an AI project), daunting, but no less essential, for which companies can no longer afford to dedicate their precious data scientists. All the more so as the volume of data is constantly increasing, especially with video: one hour of video requires 800 hours of annotation work. This is where the specialist companies come in.

‍

‍Video: Understanding Data Annotation

‍Bonus video: Understand what artificial intelligence (AI) is?

Quantity and quality

Thanks to crowdsourcing, they are able to provide very specific data with a strong guarantee of control and quality:

• the workforce is continually trained, coached and managed to stick as closely as possible to each project.
• Internal teams are constantly working on improving algorithms to speed up manual annotation processes.

An essential service for certain sectors, where data quality is vital. Like the startups that develop the autonomous vehicle like Waymo, Argo AI or Lyft, which can spend millions of dollars every month just annotating their data.

A strong business challenge

Indeed, if the quality of the data is not good, not only will the performance of the AI be unsatisfactory, but it will be necessary to start from scratch: collect and annotate the data again. This is time-consuming and costly work, which can have a decisive impact on a company's competitiveness in a context of strong competition. Hence the importance of choosing the right annotation service provider.

‍

Isahit: when technological expertise meets social impact

Isahit has developed a powerful in-house labelling tool in order to meet all the requirements and specificities of its customers. It guarantees, throughout the project :

• quality. Our workers/hiters receive customized training to familiarize themselves with each project and are subject to an integrated intelligent quality control.
• agility. Without any intermediary, our platform adapts to your projects in just a few clicks.
• security and confidentiality. Our platform and API are fully secure and our customer back office allows you to follow the project's progress live, while leaving you in control of the confidentiality and control of personal data.

But beyond this state-of-the-art technology, isahit is above all the will to have a impact positive about the world. As the first socially responsible outsourcing platform for digital tasks, we enable our guests, young women from developing countries, to become digitally emancipated. A winning approach for the latter, but also for companies in the deployment of their CSR policy.

A word from Isabelle Mashola - CEO and co-founder of isahit

"The added value of isahit, the excellence of our work, lies first and foremost in our community. Trained, committed, cultivating mutual aid, she is proof that artificial intelligence can be a vector of social and human progress #AIforHumanity."