最終更新:2021-10-08 (金) 15:51:56 (893d)
MediaPipe Face Mesh
Top / MediaPipe Face Mesh
Real-time Facial Surface Geometry from Monocular Video on Mobile GPUs
https://sites.google.com/view/perception-cv4arvr/facemesh
refine_landmarks
- Added an refine_landmarks option in the Solution APIs to further improve landmarks around eyes and lips, and output additional landmarks around the irises.
- https://google.github.io/mediapipe/solutions/face_mesh.html#attention-mesh-model
- https://github.com/google/mediapipe/releases/tag/v0.8.8
- https://twitter.com/Nextremer_nb_o/status/1445929891820490761
関連
参考
- https://ai.googleblog.com/2019/03/real-time-ar-self-expression-with.html
- https://github.com/google/mediapipe/blob/master/mediapipe/docs/face_mesh_mobile_gpu.md
実行
GLOG_logtostderr=1 bazel-bin/mediapipe/examples/desktop/hand_tracking/hand_tracking_gpu \ --calculator_graph_config_file=mediapipe/graphs/hand_tracking/hand_tracking_mobile.pbtxt