Computer Vision Platform Datagen Raises $50M Series B
Datagen, a firm specializing in computer vision artificial intelligence, announced it has raised $50 million in a Series B round bringing its total financing to $70 million.
Computer vision (CV) has quickly become ubiquitous in many industries, including retail, agriculture, manufacturing, and more, and it supports technology such as autonomous vehicles, augmented reality, and facial recognition. As a subset of AI, it allows computers to learn and understand the visual world through digital images and videos. CV applications are capable of classifying, detecting, and differentiating objects in the real world as they are trained through deep learning models.
One problem with CV technology is a shortage of sufficient training data for those deep learning models. Datagen noted findings from a commissioned report that revealed 99% of CV teams lacked the training data needed for machine learning projects, leading to their ultimate cancellation. A solution for this is using synthetic data for training, a method reportedly employed by 96% of teams in the same study. Synthetic data is data that is generated by simple rules, statistical modeling, or simulation, as opposed to manually collected real-world data, and Gartner predicts that in just a few years, 60% of data used in AI and analytics applications will be synthetic.
Specifically for CV applications, Datagen says synthetic data “obviates the need for producing or sourcing scant 2D and 3D training data for computer vision AI development.”
Datagen’s self-service platform is focused on CV teams and their synthetic data needs, and its “unique offering makes it easy for CV engineers to engage and adopt synthetic data by running Data Generation Units (DGUs) by the hour to produce the data for their AI needs. For the first time, buying synthetic data is as easy as buying cloud computing resources.”
Datagen was co-founded in 2018 by CEO Ofir Chakon and CTO Gil Elbaz, computer vision specialists who saw an opportunity for AI-powered simulations in the field. “We knew that performance increases with data in the deep learning era, and that manual data acquisition is the number one bottleneck holding the industry back from expediting orders of magnitude faster,” Chakon said in a company blog post.
The company bills itself as a “data-as-code leader” and says that data-as-code is a new paradigm in AI development, as synthetic data allows AI models to be products of pure code, enabling “data-centric AI.” This “turns heavy operational processes into a seamless, easy-to-control programmable interface,” leading to more automation and collaborative capabilities. The company says its data-centric platform allows data scientists to save critical time and resources by eliminating the need to source and manually annotate training data.
Datagen claims an 8x growth in its total revenue brought on by increased demand for its synthetic data solutions, and it says several “global tech giants” and Fortune 500 companies are using the platform for AI production needs. This Series B funding round was led by Scale Venture Partners, with participation from existing investors TLV Partners, Viola Ventures, and Spider Capital.
“The need for robust, high-variance and high-performance training data will continue to grow exponentially as computer vision algorithms and their applications become more numerous and diverse,” said Chakon. “Our mission is to enable every CV team with the best synthetic data solution to power the development of their AI applications. That’s why we are honored to welcome Scale VP to the community of our marquee investors. With today’s new funding, we are poised to accelerate growth and take the market by storm.”