Case study
VADE turns curb data into parking insight for cities, using IoT cameras, computer vision and a real-time API. Training that computer vision meant annotating enormous volumes of footage, fast and accurately.
When VADE takes on a municipal project, they work through enormous amounts of camera data in a short window, and the annotations have to be highly accurate or the resulting reports aren't worth acting on. That combination of volume, speed and accuracy is hard to staff for, and it isn't the problem a young company should be spending its own hours on.
VADE's footage needed annotators who understood the task: vehicles, spaces and events in real street scenes, labeled to the accuracy a computer vision model can actually learn from. We staffed people trained on exactly this kind of video and image work.
The team was working within 48 hours of the brief. Werkit managed it end to end: onboarding, quality checks and throughput stayed our problem, so VADE's own team stayed on their product.
Municipal projects shift. When VADE adjusted scope on short notice, the team adapted and scaled with the project instead of renegotiating it, and communication stayed prompt throughout.
The team delivered accurate, reliable annotations across millions of frames, ahead of schedule. Costs came down, the quality helped VADE win over its own clients, and when the project grew, the team grew with it.
The pattern is the same one behind our Verbit transcription programme: a managed, trained team that takes the whole queue, so the client's own people stay on the product.
Volume, accuracy and a deadline are only a crisis when the labeling team is an afterthought. A specialized team, calibrated on your data before it scales and managed by someone accountable for the whole queue, turns annotation from the bottleneck into the part that just runs.
See how our managed data teams workStart a project
Share the brief. A senior expert reads it and comes back with scope questions or a fixed-price proposal.
Prefer email? hello@werkitdata.com
