Annotated 1.2M images for a perception model
An autonomous robotics startup
A continuous semantic-segmentation pipeline at 1.2M-image scale with weekly delivery, gold-set calibration, and inter-annotator agreement tracking — production-grade training data on a release cadence.
This case study is part of our AI Annotation work — see how the same approach scales for other teams in our portfolio.
Challenge
What we walked into.
Their perception team needed clean, consistent labels at a scale and cadence that an in-house team couldn't sustain. Throughput was capping the rate at which they could iterate on the model, and quality drift between batches was hurting eval scores.
What we did
The work, step by step.
Designed a labelling ontology and gold-set in collaboration with the model team, used to calibrate every annotator before they touched real data
Ran a continuous semantic-segmentation pipeline at 1.2M-image scale with weekly delivery
Tracked inter-annotator agreement per class and shipped a QA report alongside every batch
Tuned guidelines mid-programme based on model-error analysis, closing the loop between data and predictions
Results
What it shipped.
Outcomes measured against the brief we agreed up front, not vanity metrics.
- Images labeled1.2M
- Throughput+62%
- IAA score0.93
More case studies
See all
Translation & LocalizationTelehealth app localized into 14 languages in under three weeks
We took a patient telehealth app from one English locale to fourteen in nineteen days — strict medical accuracy, full RTL support for Arabic, localized app-store listings, and a continuous-localization pipeline tied to the weekly release.
Read case study
Translation & Localization8-market e-commerce localization for a DTC brand
We localized 20k SKUs, the brand voice, the checkout, and every legal page across eight European markets — wiring transcreated campaign copy and a managed termbase into the same release as international SEO and paid marketing.
Read case study
Translation & LocalizationE-learning platform localized (video + UI) into 12 languages
Hundreds of hours of course video subtitled and dubbed, the full product UI localized, and a terminology pipeline that holds technical accuracy across twelve languages — shipped to a strict course-launch schedule.
Read case studyBereit für internationales Wachstum?
Erzählen Sie uns von Ihrem Projekt — wir melden uns innerhalb eines Werktages bei Ihnen.