Google has partnered with Figure Eight as a human-in-the-loop platform to create or help you improve the quality of your AutoML Vision datasets. Figure Eight offers:
- Human labeling of images, to give a model high quality (and large quantities of) training data
- People to review the model output to identify problems with the training dataset
- An assigned consultant to provide guidance and best practices for your labeling job design so that you can achieve the highest quality results
Request human labeling from Figure Eight
To request a human labeling job directly from Figure Eight:
Get familiar with the overall job request process by reading Figure Eight’s Data Categorization Quick Start Guide.
Create and configure your job using Figure Eight’s AutoML Image Categorization Template.
Contact Figure Eight for your initial consultation.
Provide Figure Eight with instructions for the labeling task.
Include as much detail as possible. Make sure to include a “None_of_the_above” category to track the images that cannot be labeled.
Launch your job.
Results will start coming in quickly and you’ll be able to monitor your results within the Figure Eight platform in real time.
Using Figure Eight data in AutoML
After your Figure Eight human labeling job completes, you will receive an email with a link to your results.