Alegion Control - A Behind the Scenes of This Powerful New Tool
Over the past year Alegion has taken on the challenge of building the best possible video annotation experience. At the same time, we were witnessing an explosion of research and focus around video data in the computer vision (CV) space. Having assessed the tooling available on the market, we knew there just had to be a better way.
We kept asking ourselves, "Can we save our customers time and money with a self-serve resource for ML teams? A solution that they can manage themselves and that produces quality training data for video annotation in a fraction of the time?”
Turns out, we could! The recent launch of Alegion Control, our first ever self-serve video annotation solution, and its consequential success, has gotten us thinking about what it took to get here.
Chip Ray, our CTO, decided to take a look back at the technical and design challenges we faced in the development of Alegion Control.
In this three part series, Chip looks at (and then explains how we solved for) :
- The initial challenges of video annotation due to size of data and sheer number of annotations and contextual information
- The scalability challenges required to support 100K+ annotations in a single video and the back-end challenges of supporting thousands of simultaneous annotators
- The user experience of our platform including playback and real-time streaming, i.e. optimizing browser performance.
Our goal from the beginning was to figure out a way to give machine learning (ML) teams direct access to self-manage their annotation process on our powerful data labeling platform.
It was this big vision that powered us through the long year of development, even when we hit complex challenges.
Ultimately, the story of Alegion Control is a story about a search for excellence and efficiency.
Read Part 1 here for the behind the scenes history of Alegion Control.