Frequently Asked Questions
How long does it take to get my training data?
For simpler labels such as bounding box,
we typically process over a million labels per day.
For more complex labels like segmentation masks, we can process over a
hundred thousand labels per day.
Contact us, and we can provide you with a guaranteed end date.
How can I see the progress of my project?
Once we start on your project, we create an account for you with which you can access the Dashboard. There you can monitor our progress and give feedback in real time. We are planning to open up the Dashboard to everyone soon. Stayed tuned!
I have my own raw data that needs to be labeled. How do I upload them?
Once we issue an account for you, you can upload the raw data on our Dashboard. For larger projects, we can arrange an interface that is most convenient for you (i.e. command-line API, etc.).
How do I receive or download the completed work?
Similar to the upload feature, the completed work can be downloaded through our Dashboard. For larger projects, we can arrange an interface that is most convenient for you (i.e. command-line API, etc.).
In what data format is the final training data delivered?
When discussing the data specifications with us, we will ask you what data format works best for you. You can choose from one of the standard JSON formats (COCO, Pascal, etc.), use your custom JSON format, or readily usable TF Records, and so on.
How does the data collection service work?
We have a very large collection of copyright-free data we have collected through partners, professional collectors and crawling. In addition, we operate a pool of crowdsourced users that can gather the data you specify. Our curated list is getting larger every minute!