6

There seems to be a lot of overlap between these 3 Google libraries.

According to their sites:

  • MediaPipe: MediaPipe offers cross-platform, customizable ML solutions for live and streaming media.

  • ARCore: With ARCore, build new augmented reality experiences that seamlessly blend the digital and physical worlds.

  • MLKit Vision: Video and image analysis APIs to label images and detect barcodes, text, faces, and objects.

Could someone with experience working with these explain how they relate to eachother and what are the use cases for each?

For example which would be appropriate to implement high level, popular features such as face filters?

(Also perhaps some insight on which of the 3 is most likely to land in Google Graveyard the fastest)

softcode
  • 4,358
  • 12
  • 41
  • 68

1 Answers1

7

Some simplified & informal explanations:

MediaPipe is a powerful but lower-level library for live and streaming ML solutions, which requires non-trivial setup and customization before it works for your use case.

ML Kit is an end-to-end solution provider, offering mobile friendly, easy-to-use APIs and pre-built pipelines under the hood. Several ML Kit features are actually powered by MediaPipe internally (i.e. Pose detection and Selfie-segmentation).

There is no direct relationships between ARCore and ML Kit, but there could be shared or similar ML models in between, because both require ML models to power their features but the two products have different focuses.

Shiyu
  • 875
  • 4
  • 5