Mediapipe

* MediaPipe Hand
* [Updated model, documentation and pipeline](https://solutions.mediapipe.dev/hands)
* Improved hand tracking quality in various scenarios
* Added handedness prediction, e.g. left or right hand.
* [See tweet for visualization](https://twitter.com/GoogleAI/status/1265319835283537921)
* MediaPipe Face Detection
* [Updated model for front-facing camera](https://solutions.mediapipe.dev/face_detection#resources)
* [Added model for back-facing camera](https://solutions.mediapipe.dev/face_detection#resources), which better handles smaller faces
* MediaPipe AutoFlip
* [Updated to use face detection model for back-facing camera](https://solutions.mediapipe.dev/autoflip)
* [Support for running MediaPipe on Windows - experimental](https://google.github.io/mediapipe/getting_started/install.html#installing-on-windows)
* MediaPipe TFLite CPU desktop live and offline demos only. No TFLite GPU and TensorFlow support yet.
* Compiled by MSVC through Bazel. No MinGW and Cygwin support yet.
* MediaPipe Android apps are NOT buildable on native Windows yet.
* [Tracing and profiling MediaPipe graphs - experimental](https://google.github.io/mediapipe/getting_started/install.html#installing-on-windows)
* [Support for collecting trace logs on Android, iOS, Linux](https://google.github.io/mediapipe/tools/tracing_and_profiling.html#collecting-the-logs)
* To analyze the trace logs, [MediaPipe Visualizer](https://viz.mediapipe.dev)
* Community related
* [Awesome MediaPipe: curation of code related to MediaPipe](https://mediapipe.org)
* [Slack community for MediaPipe users](https://mediapipe.page.link/joinslack)

* [MediaPipe KNIFT](http://solutions.mediapipe.dev/knift)
* See also [Google Developers Blog post](https://developers.googleblog.com/2020/04/mediapipe-knift-template-based-feature-matching.html)

* [MediaPipe Face Mesh](https://solutions.mediapipe.dev/face_mesh) (3D face landmark estimation with 468 landmarks)
* [MediaPipe Modules](https://github.com/google/mediapipe/tree/master/mediapipe/modules)
* [Face Detection](https://github.com/google/mediapipe/tree/master/mediapipe/modules/face_detection)
* [Face Landmark example](https://github.com/google/mediapipe/tree/master/mediapipe/modules/face_landmark)

* Support Bazel 2 and 3. MediaPipe will no longer support Bazel 0.x and 1.x.
* MediaPipe's TensorFlow dependency is now pointing to https://github.com/tensorflow/tensorflow/commit/805e47cea96c7e8c6fccf494d40a2392dc99fdd8, which is submitted on April 1, 2020.
* MediaPipe's Protobuf dependency is now pointing to [protobuf-3.11.4](https://github.com/protocolbuffers/protobuf/releases/tag/v3.11.4).
* Solve Xlib's Status & absl::Status compilation issue on linux.
* Fix setup_opencv.sh.
* Fix media sequence documentation by following the suggestions in https://github.com/google/mediapipe/issues/566.

* CPU acceleration with [XNNPack](https://github.com/google/XNNPACK)
* Fix iOS memory leak issue: https://github.com/google/mediapipe/issues/387
* Support detection proto in MediaPipe AAR
* Add hair segmentation desktop live demo
* [Google Open Source code search for MediaPipe Github repository](https://cs.opensource.google/mediapipe/mediapipe)

* [MediaPipe Objectron](https://solutions.mediapipe.dev/objectron) (3D object detection and tracking using only RGB images)
* [Objectron calculators](https://github.com/google/mediapipe/tree/master/mediapipe/graphs/object_detection_3d/calculators)
* See also [Google AI Blog post](https://ai.googleblog.com/2020/03/real-time-3d-object-detection-on-mobile.html)
* Bug fixes
* Update Coral dependences to match v0.6.9 release