Cord – a start-up company automating annotation for computer vision data – has raised a $12.5M Series A round, a fast follow on to a $4.5M seed round in June of this year. The round is led by CRV with participation from Y Combinator Continuity, Harpoon Ventures, and Crane Venture Partners.
Read More: Interworks.Cloud Discusses With HPE About The Evolution Of Cloud Services In Recent Webinar
“We knew early on that we needed a new technologically-focused method to solve this problem”
Co-Founders Ulrik Stig Hansen and Eric Landau realized that the next generation of AI applications will be driven by data-centric approaches centered around training data creation, management, and evaluation.
Current methods are reliant on workforces of millions of people to prepare training datasets for AI use. They created Cord to replace these manual processes that make AI development expensive, time consuming, and difficult to scale.
“We knew early on that we needed a new technologically-focused method to solve this problem,” said Hansen. “We realized that rather than automating the annotation process with one single monolithic model, we could break down the problem into very small and manageable components, each automated with their own ‘micro-model’. Basically, we have reduced the entire labelling process to an automated assembly line fuelled by our revolutionary micro-model approach”.
Cord’s meteoric growth in less than a year, and new micro-model approach in a very competitive market is a testament to the founders technical strength, and grit. There is no doubt that businesses need a more scalable way to annotate data, and Cord is the scalable solution.
Read More: Apperio Included As A Representative Vendor In 2021 Gartner Market Guide For Corporate Legal Spend…
Since the company’s launch in early 2021 and founding in 2020, the platform has facilitated the annotation of more than 100 million frames and images and served customers in verticals such as medical imaging, smart cities, sports analytics, satellite imaging, and more. Cord’s approach has allowed customers to build their own custom automated annotation pipelines and workflows, and has dramatically reduced the time and cost to get their models developed. Usage is scaling rapidly, with the number of labels generated on the platform growing ~70% month-over-month.
Cord will continue to scale the organization with plans to expand its set of automation tools, data curation features, and model evaluation modules. “We have made a lot of progress so far, but we still have a long path ahead of us. Our team is working tirelessly to continue building a platform that allows for labelled training data to be created largely by machines rather than by humans,” said Hansen.