1. What's your business model?
We have two business models that intersect depending on who owns the Intellectual Property of the images/videos being annotated. If we're annotating your data you retain all IP rights and we operate a Software as a Service business model. We also have our own datasets which are available under a licensing model, in this agreement we retain all IP rights to the data.
2. Where is your workforce located?
Our headquarters is in Silicon Valley, and most management and annotators are in Eastern Europe. We have also hired from the Philippines and Mexico to provide overnight coverage.
3. Do you crowdsource?
How do you hire your workers? We do not crowdsource our annotation team. They are a highly trained, full-time, professional workforce. However we do crowdsource the photos and videos taken for our own datasets. These are collected using our Mobile app on the Google Play Store.
4. How big is your workforce?
We have 100 full-time annotators and 8000 crowdsourced photographers (with 1000 daily active users).
5. What does your annotation process look like?
Our primary focus is on the quality and accuracy of results, so we have devised a multi-step process that combines both humans and machine learning models. Firstly humans create annotations with the assistance of AI. These annotations are then checked for quality assurance by another human layer. Finally, we have final checks by both human and AI at a dataset and model level too.
6. How can we track annotation progress?
You are given access to a console where you have 24/7 access to your real-time results as soon as they've gone past our first level of QA.
7. How do we provide feedback? What if something is not right?
On your first contact with us you will be assigned a dedicated Data Concierge whose job it is to get you the results you need as efficiently as possible. We won't start the full-rate production until we've reached a full agreement with you first via trial tasks, regular meetings and feedback provided via your console.
8. Are there volume discounts?
Yes, we offer generous discounts for upfront payment and volume. Please contact us for details.
9. What is the results output format?
As you prefer, typically CSV or JSON.
10. What do you charge?
Prices depend on the complexity of the taxonomy and if you require additional annotations such as attributes. Typically bounding boxes cost from 2c - 4c each and segmentations from 10c - 20c each.
11. What's the turnaround time?
For a first batch of 50,000 images its usually 2-3 weeks.
12. How quickly can you scale?
25 annotators per week.