CVAT began as an open-source program at Intel in 2017 to accelerate the annotation of digital images and videos for training computer vision algorithms. Back then, our mission was to help internal data science teams obtain new annotated data to train deep neural networks. Little did we know that, eight years later, we’d be helping hundreds of thousands of teams build better AI models.
Three years ago, we spun out into an independent company to build on our leadership position in visual data annotation for computer vision and machine learning applications. Since then, we’ve achieved some impressive feats as a team and as a business. One of them is our vibrant developer community on GitHub, which has helped shape CVAT’s roadmap and core architecture in powerful ways.
To celebrate our three-year anniversary and 14,000 stars on the GitHub repository that started it all, we’re sharing 14+ key (and fun) milestones that brought us here:
Milestone #1: January 2017
Intel software engineers Nikita Manovich (Team Lead of Data Infrastructure) and Andrey Zhavoronkov (Software Engineer) begin developing an internal annotation tool by enhancing the VATIC tool. They add image annotation, attribute support, and redesign the client-server architecture. Milestone #2: June 2018
The team decides to go open-source. Since the team also supported OpenCV, a popular computer vision library, they released the first version of CVAT (Computer Vision Annotation Tool), 0.1.0, on GitHub under the same organization. Because of that, the project quickly gained traction. Milestone #3: December 2019
Intel officially announces CVAT as its new open-source initiative to streamline digital image and video annotation for computer vision use cases. Milestone #4: September 2020
The CVAT open-source community grows rapidly, contributing new features, fixes, and integrations. We also launch cvat.org (no longer available) – a free online server for annotating data without installing CVAT locally. Milestone #5: April 2021
The GitHub project becomes one of the most-starred in its category, reaching 5,000 stars by late 2021. CVAT is also included in the GitHub Arctic Code Vault, preserving our code in a long-term Arctic archive of notable open-source projects. Milestone #6: February 2022
CVAT partners with HUMAN Protocol, a decentralized labor marketplace, to allow customers to scale annotation workflows with on-demand, crowdsourced annotators. Milestone #7: July 2022
CVAT officially spins out from Intel and becomes an independent company. Nikita Manovich and Boris Sekachev, who joined Intel as an intern during summer of 2017, become co-founders, ready to take CVAT to the next level. Milestone #8: December 2022
We launch CVAT Online—a cloud-hosted version of our open-source platform, giving teams and individuals access to scalable, collaborative annotation workflows without needing to self-host. Milestone #9: August 2023
CVAT Online reaches 50,000 users. We also sign our first enterprise contract, laying the foundation for CVAT Enterprise—a commercial version of our self-hosted annotation platform with support for automation, team workflows, and enterprise-grade integrations. Milestone #10: September 2023
We reach 10,000 stars on GitHub and secure our first paid labeling contract, launching our Labeling Services business – a natural extension of our mission to provide end-to-end annotation solutions. Milestone #11: February 2024
CVAT joins Google Summer of Code 2024, enabling students and new contributors around the world to work on real-world annotation challenges. Milestone #12: April 2024
We introduce annual plans for CVAT Online, helping our customers save up to 30% on premium features. Transparent, flexible pricing becomes a core part of our user experience, especially for teams with long-term annotation pipelines. Milestone #13: May 2024
CVAT’s labeling services scale to a global workforce of several hundred annotators. Our clients now include Fortune 100 companies across retail, logistics, robotics, and more. We’re also honored to be recognized as a top-choice annotation tool at Embedded Vision Summit 2024. Milestone #14: October 2024
We introduce SAM2-powered video tracking in CVAT Enterprise, enabling video annotation to be completed up to 10x faster than before – a massive leap in productivity for teams working with surveillance, autonomous driving, and motion datasets. Milestone #15: January 2025
AI agents come to CVAT. Customers can now integrate their own detection and segmentation models to automate labeling across CVAT Online and CVAT Enterprise. Milestone #16: June 2025
CVAT Analytics expands: teams can now track annotation efficiency, reviewer throughput, rework rates, and annotator performance trends in real time. This helps customers make data-driven decisions about scaling their workforce and improving dataset quality over time. What’s Next?
As more companies embed AI into their operations, from warehouses and self-driving fleets to retail shelves and hospitals, supervised learning will remain the backbone of innovation. And supervised learning requires high-quality, structured annotated data.
In our fourth year as an independent company, we’re more focused than ever on giving teams everything they need to produce that data fast, accurately, and at scale. We’re not just building an annotation platform. We’re building the foundation for better AI.
Thank you for being part of the journey!


Company News
July 24, 2025
CVAT Celebrates 14K Stars on GitHub and Its Three-Year Anniversary!