• 本文
• リロード
• 差分
• バックアップ
• 名前変更
• リファラ
• 新規ページ
• 編集

Google

MediaPipe

Top / MediaPipe

MediaPipe is the simplest way for researchers and developers to build world-class ML solutions and applications for mobile, desktop/cloud, web and IoT devices.

a framework for building multimodal (eg. video, audio, any time series data), cross platform (i.e Android, iOS, web, edge devices) applied ML pipelines

マルチモーダル(ビデオ、オーディオ、時系列データなど)を利用したクロスプラットフォームのMLパイプラインを構築するためのフレームワーク

https://mediapipe.dev/

https://github.com/google/mediapipe

https://google.github.io/mediapipe/

ビジュアライザ

• https://viz.mediapipe.dev/

ML Solutions in MediaPipe

• Face Detection (BlazeFace)
• MediaPipe Face Mesh
• MediaPipe Iris
• Hand Detection
• Hand Tracking
• Multi-hand Tracking
• Hair Segmentation
• Object Detection
• Object Detection and Tracking
• Objectron: 3D Object Detection and Tracking
• AutoFlip?: Intelligent Video Reframing
• KNIFT: Template Matching with Neural Image Features

モデル

• https://github.com/google/mediapipe/tree/master/mediapipe/models

• object_detection_saved_model?
• face_detection_front.tflite?
  • BlazeFace

• face_detection_front_labelmap.txt?
• face_landmark.tflite?
  • Face Mesh
• hair_segmentation.tflite?
• hand_landmark.tflite?
• hand_landmark_3d.tflite?
• knift_float.tflite?
• knift_float_400.tflite?
• knift_index.pb?
• knift_labelmap.txt?
• object_detection_3d_chair.tflite?
• object_detection_3d_sneakers.tflite?
• palm_detection.tflite?
• palm_detection_labelmap.txt?
• ssdlite_object_detection.tflite?
• ssdlite_object_detection_labelmap.txt?

GPU

• MediaPipe framework doesn’t require CUDA for GPU compute and rendering. However, MediaPipe can work with TensorFlow to perform GPU inference on video cards that support CUDA.

関連

• TensorFlow
• TFLite?

入門

• https://note.com/npaka/n/n5008f82c43cc