Video and image annotation tools built for scale
Short history of Keylabs
Our story began as an image and video annotation service provider. Being perfectionists in what we do, we were not satisfied with existing industry tools that would allow us to work fast and efficiently, especially when annotating video and large datasets. Many tools on the market were good... but not enough. So we decided to build our own toolset that would satisfy any inner perfectionist. There were many different challenges on the way, ranging from project management flow to data processing speed to data security to scalable storage, and we were able to resolve them all. We took our years of experience and innovative thinking, combined it with productivity and data security, wrapped it into user-friendly design and got the best product on the market. Meet Keylabs, our all-in-one annotation platform.
Been using this tool for a few months. Things are pretty intuitive, there is nothing that takes more clicks to use than it should. It's definitely designed with the user in mind, by someone familiar with annotation work, the interface is not detached from the process. Now that I learned to combine fast automatic and precise manual work, I'm way more productive now.
We used the Keylabs tool to run a set of labeling streams and was impressed with the project management aspect of it. There's a lot of nifty features to monitor how well you are doing in terms of workforce and task completion, seems like a lot of thought went into those.
Video annotation feature has been a lifesaver. We process a lot of HD video, and the system is pulling its weight pretty nicely, object tracking works smoothly, and process times are reasonable.
Tried a few tools and Keylabs came out on top - the only one that had the features we need. We wanted something that allows meaningful collaboration on the same instances, in real time. Another important part is that some validation workflow is hardwired into the tool and is configurable - our QA process is very specific. Both worked well for us.