The domain is "Training Data".
It's one integrated system that handles everything in the new Training Data abstractions world.
It integrates with adjacent tools.
The largest group of end users is data annotators and subject matter experts.
A data scientist or similar sets up the Schema (labels and attributes), the dataset structure, etc.
An admin or project manager sets up the human annotation pipelines.
A software engineer sets up the overall system, data permissions, and maintaining a large deployment.
Diffgram is a drop in replacement for most of the functions of the following systems:
Labelbox, CVAT, SuperAnnotate, Label Studio (Heartex), V7 Labs (Darwin), BasicAI, SuperbAI, Kili-Technology, HastyAI, Dataloop, Keymakr.
Just ask for it and send them this link: https://github.com/diffgram/diffgram. They can use the open source version in many use cases without additional cost. They are welcome to look at https://diffgram.com/enterprise for larger cases. If you are purchasing or using the Enterprise license you can naturally use any team you want. Examples for firms you can "ASK FOR DIFFGRAM" (Appen, Lionbridge, Cloud Factory, SixGill, iMerit)
A Diffgram instance has multiple services (online, batch processing, router, database, and BLOB storage) - but that abstraction is generally hidden. Within the training data realm Diffgram covers the majority of use cases.
Open Source: LICENSE
Enterprise: Custom. Please contact us.
Updated about 1 month ago