Voxel51 is a company that provides open-source computer vision tools for machine learning, including FiftyOne, a toolkit that helps improve the quality of datasets and deliver insights about models.
Detailed Background Story
Voxel51 was founded in 2017 by Jason Corso, a professor of electrical engineering and computer science at the University of Michigan, and Brian Moore, a former student of Corso’s. The company’s name comes from the idea of a “voxel,” a three-dimensional pixel used in computer graphics and imaging.
The founders of Voxel51 saw a need for better tools to manage and analyze large datasets in computer vision, particularly in the field of machine learning. They believed that many companies and researchers were struggling to keep up with the volume and complexity of data being generated, and that this was hindering progress in the field.
To address this problem, Voxel51 developed FiftyOne, an open source toolkit that enables users to curate better data and build better models. FiftyOne includes tools for data exploration, visualization, and management, as well as features for collaboration and sharing.
Today, Voxel51 is a growing company with a team of engineers, data scientists, and other experts working to advance the field of computer vision and machine learning. The company is committed to open source development and collaboration, and its tools are used by tens of thousands of engineers and researchers around the world.
Voxel51’s target customers are developers, data scientists, and researchers who work with computer vision and machine learning. Their open source toolkit, FiftyOne, is designed to help these professionals improve the quality of their datasets and deliver insights about their models. Additionally, FiftyOne Teams is designed for teams working on computer vision projects who need to securely collaborate on datasets in the cloud. Voxel51’s products are used by a wide range of industries, including automotive, healthcare, and robotics, among others.
Capital Raised, Estimated Revenue
Voxel51 has raised $15.8 million to date, as of the time of the article’s publication. As the primary product revenue is yielded through its collaboration tool for managing datasets, it is not clear how substantial the revenue is to date.
Products and Services
- Open source computer vision tools for machine learning
- FiftyOne toolkit to improve dataset quality and deliver insights about models
- Integration with popular machine learning tools and public datasets like COCO, Open Images, and ActivityNet
- Ability to create custom datasets from scratch
- FiftyOne Teams, a fully managed version of FiftyOne with cloud-backed features for teams
- Consulting services for computer vision and machine learning projects
There are several companies that offer similar products and services to Voxel51. Some of Voxel51’s competitors in the computer vision and machine learning space include:
Pros and Cons of Voxel51
- Open source: Voxel51’s tools are open source, meaning that users can access and modify the code to suit their needs.
- Integrations: FiftyOne integrates with popular machine learning tools and public datasets, making it easier for users to work with the tools they’re already familiar with.
- Customizable datasets: FiftyOne allows users to create their own datasets from scratch, giving them more control over the data they’re working with.
- Insights: FiftyOne provides insights about models, helping users to better understand how their models are performing and identify areas for improvement.
- Fully managed option: Voxel51 offers FiftyOne Teams, a fully managed version of FiftyOne that adds cloud-backed features specifically built for teams.
- Consulting services: Voxel51 provides consulting services to help organizations with their computer vision and machine learning projects, which could be helpful for those who need additional support.
- Voxel51 is built by researchers for researchers. While the features for tagging data and managing metadata are robust, the interface is clunky. There is a learning curve for being able to leverage the full feature set.
- Voxel51 does integrate with several toolkits, but there is no internal service or application for labeling image data. This is a major gap compared to alternative solutions.