Mediapipe

* MediaPipe Multi-hand tracking example app
* For [Android](https://github.com/google/mediapipe/tree/master/mediapipe/examples/android/src/java/com/google/mediapipe/apps/multihandtrackinggpu), [iOS](https://github.com/google/mediapipe/tree/master/mediapipe/examples/ios/multihandtrackinggpu) GPU, [desktop CPU/GPU](https://github.com/google/mediapipe/tree/master/mediapipe/examples/desktop/multi_hand_tracking)
* [BeginLoop](https://github.com/google/mediapipe/tree/master/mediapipe/calculators/core/begin_loop_calculator.h)/[EndLoop calculators](https://github.com/google/mediapipe/tree/master/mediapipe/calculators/core/end_loop_calculator.h) for processing a collection of elements iteratively
* Updated dependencies
* Support Bazel 1.1.0
* Update TensorFlow dependency to the most recent version
* Use ‘-std=c++14’ by default
* Use Google absl’s flat_hash_set and flat_hash_map for better performance.
* Bug fixes
* Fix a bug in OpenCVVideoDecoderCalculator for [issue 204](https://github.com/google/mediapipe/issues/204)
* Fix a bug in kinetics_dataset.py for [issue 228](https://github.com/google/mediapipe/issues/228)
* Documentation
* Multi-hands [mobile](https://github.com/google/mediapipe/blob/master/mediapipe/docs/multi_hand_tracking_mobile_gpu.md) & [desktop](https://github.com/google/mediapipe/blob/master/mediapipe/docs/multi_hand_tracking_desktop.md)
* Updated [list of public events for MediaPipe](https://github.com/google/mediapipe/blob/master/README.md)

* [MediaPipe Android development with Gradle](https://github.com/google/mediapipe/blob/master/mediapipe/docs/android_archive_library.md)
* Building Android AAR using Bazel
* Using MediaPipe built AAR in Android Studio
* [YT8mil baseline model inference pipeline](https://github.com/google/mediapipe/blob/master/mediapipe/docs/youtube_8m.md#model-inference-for-youtube-8m-challenge)
* Bazel versions supported are v0.24.1 to v0.29.1
* Bug fixes
* Android AAR for [issue 88](https://github.com/google/mediapipe/issues/88)
* Documentation
* [Model card for YT8mil baseline model](https://drive.google.com/file/d/1xTCi9-Nm9dt2KIk8WR0dDFrIssWawyXy/view)
* MediaPipe Android development with Gradle
* YT8M model inference

* [MediaPipe CPU & GPU graph runner for Desktop graphs with webcam](https://github.com/google/mediapipe/blob/master/mediapipe/examples/desktop/demo_run_graph_main.cc)
- Mobile graphs example & doc
- [Face Detection](https://github.com/google/mediapipe/blob/master/mediapipe/docs/face_detection_mobile_cpu.md)
- Mobile CPU
* Desktop graph examples & docs
- [Face detection](https://github.com/google/mediapipe/blob/master/mediapipe/docs/face_detection_desktop.md)
- Desktop GPU
- Desktop CPU
- [Object detection](https://github.com/google/mediapipe/blob/master/mediapipe/docs/object_detection_desktop.md)
- Desktop CPU
- [Hand tracking](https://github.com/google/mediapipe/blob/master/mediapipe/docs/hand_tracking_desktop.md)
- Desktop GPU
- Desktop CPU
- [Hair segmentation](https://github.com/google/mediapipe/blob/master/mediapipe/docs/hair_segmentation_desktop.md)
- Desktop GPU
* Allows android to print log output from native code. Fixed [issue 118](https://github.com/google/mediapipe/issues/118)
* Fix bug in YT8m feature extraction pipeline

* New Features
- [YT8m Feature extractor](https://github.com/google/youtube-8m/tree/master/feature_extractor) for [Kaggle challenge](https://www.kaggle.com/c/youtube8m-2019) for [ICCV workshop](https://research.google.com/youtube8m/workshop2019/index.html)
- [Deepmind Kinetics I3D](https://github.com/deepmind/kinetics-i3d) [pre-processing pipeline](https://github.com/google/mediapipe/blob/master/mediapipe/examples/desktop/media_sequence/kinetics_dataset.py)
* Bug fixes
- OpenCVVideoDecoderCalculator for [issue 57](https://github.com/google/mediapipe/issues/57)
* Documentation
- Fix to Android studio setup instruction for [issue 65](https://github.com/google/mediapipe/issues/65)
- Fix to iOS setup bundle ID for [issue 50](https://github.com/google/mediapipe/issues/50)
- Fix to Android NDK location for [issue 98](https://github.com/google/mediapipe/issues/98)

* Subgraph API in C++
* Graph Execution API in Obj-C
* [Updated previous mobile examples with iOS](https://github.com/google/mediapipe/blob/master/mediapipe/docs/examples.md)
- Edge detection with GPU (Hello World! on iOS )
- Object Detection with GPU
- Face Detection with GPU
- Face Detection with CPU
* 2 new mobile examples with Android & iOS
- [Hand detection with GPU](https://github.com/google/mediapipe/blob/master/mediapipe/docs/hand_detection_mobile_gpu.md)
- [Hand tracking with GPU](https://github.com/google/mediapipe/blob/master/mediapipe/docs/hand_tracking_mobile_gpu.md)
* [A new data preprocessing graph for preparing kinetics dataset](https://github.com/google/mediapipe/blob/master/mediapipe/examples/desktop/media_sequence/kinetics_dataset.py)
* [Audio Calculators - decoding, resampling, spectrogram, mfcc](https://github.com/google/mediapipe/tree/master/mediapipe/calculators/audio)
* [Visualizer](https://viz.mediapipe.dev)
- Multi-tab support
- Subgraph visualization
* [Documentation](https://github.com/google/mediapipe/tree/master/mediapipe/docs)
- Added iOS section for edge detection, face detection, and object detection
- [Google AI blog post on hand tracking](https://mediapipe.page.link/handgoogleaiblog)
- Hand detection
- Hand tracking
- [Hand model card](https://mediapipe.page.link/handmc)
- Subgraph visualization

Major Features
* Calculator API in C++
* Graph Construction API in ProtoBuf
* OpenGL ES support for Calculator API
* Graph Execution API in C++
* Graph Execution API in Java (Android)
* 5 Desktop and Android examples (Graphs, Calculators, TF/TFlite models)
- Edge detection on GPU for Android (Hello World! on Android)
- Object Detection on CPU and GPU for Android
- Face Detection on GPU for Android
- Hair Segmentation on GPU for Android
- Object Detection on Desktop (No desktop GPU support)
* Visualizer launched at https://viz.mediapipe.dev
* Dataset preprocessing using MediaSequence
* Set of 59 calculators for core framework, image, tensorflow, tflite, util, video
* Initial v0.5 documentation launched at https://docs.mediapipe.dev