V7 raises $33 million to automate training data for computer vision AI models

V7 raises $33 million to automate training data for computer vision AI models

AI promises to help, and maybe even replace, humans accomplish everyday tasks and solve problems that humans haven’t been able to address, but ironically, building that AI faces a major scalability problem . It is only as good as the models and data used to train it, so increasingly large collections of data need to be sourced and imported. But logging and manipulating that training data takes a lot of time and money, slowing down work or overall effectiveness, and perhaps both.

A startup called V7 Labs believes it has taken a step forward in how this is addressed. They are training models that are effectively built to automate the training of those models. Today it announces $33 million in funding to fuel its growth after seeing high demand for its services.

V7’s focus today is on computer vision and helping to identify objects. She says she can learn what to do from just 100 human-annotated examples.

It currently has strong traction in the fields of medicine and science, where its platform is being used to help train AI models to speed up, for example, how cancers and other problems are identified in scans. V7 is also starting to see activity with technology companies and technology experts looking at how to apply its technology in a wide variety of other applications, including companies building engines to create images from natural language commands and industrial applications. He’s not revealing a complete list of customers and those evaluating his technology, but the list has more than 300 customers and includes GE Healthcare, Paige AI and Siemens, along with other Fortune 500 companies and large private companies.

Radical Ventures and Temasek are co-leaders of this round, with Air Street Capital, Amadeus Capital Partners and Partech (three previous backers) also participating, along with a number of leading people in the world of machine learning and intelligence artificial. They include Francois Chollet (the creator of Keras, the open source Python neural network library), Oriol Vinyals (a principal researcher at DeepMind), Jose Valim (creator of the Elixir programming language), Ashish Vaswani (a co-founder of Adept AI who was previously at Google Brain, where he invented Transformers) and unnamed others from OpenAI, Twitter, and Amazon.

Chief Executive Alberto Rizzoli said in an interview that this is the largest Series A funding round in this category to date, and will be used to both hire more engineers and grow its business operations to address a new surge of interest from clients with an emphasis on the U.S. He declined to comment on the valuation, but the startup has now raised about $36 million, and from what I understand the valuation is now around $200 million.

Rizzoli also declined to talk about revenue figures, but said ARR tripled in 2022

A number of other startups have emerged to help improve the efficiency of training AI data and to address the broader area of ​​AI modeling. SuperAnnotate, which raised about $18 million for PitchBook, is one of V7’s closest competitors. (V7 also illustrates how the two services compare.) Others include Scale AI, which initially focused on the automotive sector but has since branched out into a number of other areas and is now valued at around $7 billion; Labelbox, which works with companies like Google and others on AI labeling; and Hive, which is now valued at around $2 billion.

V7 – named in reference to the fact that AI is the “seventh” area for image processing after the six areas of the human brain that form its visual cortex (V1 to V6) – and the others point to the fact that the training model is inefficient and can be improved.

The specific USP of V7 is automation. It is estimated that about 80% of an engineering team’s time is spent managing training data: tagging it, identifying when something is mislabelled, rethinking categorizations, and so on, and then built a model for automate this process.

He calls the process he devised “programmatic tagging”: Using generic AI and his own algorithms to segment and label images, Rizzoli (who co-founded the company with his CTO Simon Edwardsson) says it only takes 100″ people-driven” for its automatic labeling to spring into action.

Investors are betting that shortening the time between conceiving and applying AI models will bring more business for the company. “Computer vision is being deployed at scale across all industries, delivering innovation and breakthroughs and a rapidly growing $50 billion market. Our thesis for V7 is that the breadth of applications and the speed with which new products should be brought to market require a centralized platform that connects AI models, code and humans in a loop ecosystem.” Pierre Socha, partner at Amadeus Capital Partners, said in a statement.

V7 describes the process as “autopilot,” but the co-pilot could be more precise: the idea is that anything marked as unclear is redirected back to humans to be evaluated and reviewed. It doesn’t so much replace those humans as it makes it easier for them to get through workloads more efficiently. (It can also sometimes perform better than humans, so the two used in tandem might be useful for double-checking each other’s work.) Below is an example of how image training is working on a lunch scan for detect pneumonia.

Image credits: labs v7

Considering the many areas where AI is being applied to improve the way images are processed and used, Rizzoli said the decision to initially double down on the field of medicine was in part to keep the startup grounded. and focus on a market that may have never built this type of technology in-house, but would definitely like to use it.

“We decided to focus on verticals that are already commercializing AI-powered applications or where a lot of work is done on visual processing, but by humans,” he said. “We didn’t want to be tied down to moonshots or projects that are depleting big R&D budgets because that means someone is trying to completely solve the problem on their own, and they’re doing something more specialized, and they might want to have their own own technology, not that of third parties like us”.

And in addition to companies searching for “their own secret sauce,” sometimes projects may never see the light of day outside the lab, Rizzoli added. “Instead, we are working for real applications,” she said.

Image credits: V7 labs (Opens in a new window)

In another aspect, startup represents a shift that we are seeing in the way information is sourced and adopted across enterprises. Investors think the framework V7 is building could potentially change the way data is captured by such firms in the future.

“V7 is well positioned to become the industry standard for data management in modern AI workflows,” Parasvil Patel, partner at Radical Ventures, said in a statement. Paten will join the V7 board this round.

“The number of problems that can now be solved with AI is vast and growing rapidly. As companies of all sizes rush to seize these opportunities, they need cutting-edge data and modeling infrastructure to deliver exceptional products that continually improve and adapt to real-world needs,” added Nathan Benaich of Air Street Capital , in a statement. “This is where V7’s AI Data Engine shines. Regardless of industry or application, customers rely on V7 to deliver robust AI-first products faster than ever before. V7 packs best practices into rapid evolution of the industry in multiplayer workflows, from data to model to product”.

#raises #million #automate #training #data #computer #vision #models

Leave a Reply

Your email address will not be published. Required fields are marked *