Skip to content
/ pose Public

Real-time Hand Pose Recognition and Dataset Generation With Mediapipe

Notifications You must be signed in to change notification settings

sofman98/pose

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Real-time Hand Pose Recognition and Dataset Generation With Mediapipe

NOTE: The GPU was used to predict and generate the dataset because it is much faster, the CPU version was almost unusable, a GPU is needed.

INSTALLATION:

HOW TO GENERATE THE DATASET:

  • cd into the installation folder: $cd mediapipe

  • Build the program with: $bazel build -c opt --copt -DMESA_EGL_NO_X11_HEADERS mediapipe/examples/desktop/hand_tracking:hand_tracking_out_gpu

  • Run the program: $bazel-bin/mediapipe/examples/desktop/hand_tracking/hand_tracking_out_gpu -- calculator_graph_config_file=mediapipe/graphs/hand_tracking/hand_tracking_desktop_live.pbtxt

  • Run Landmarks_classification/dataset_managing.py in order to process the data

HOW TO TRAIN SVM:

  • Run Landmarks_classification/train.py

HOW TO PREDICT:

  • Build the program with: $bazel build -c opt --copt -DMESA_EGL_NO_X11_HEADERS mediapipe/examples/desktop/hand_tracking:hand_tracking_predict_gpu

  • Run the program: $bazel-bin/mediapipe/examples/desktop/hand_tracking/hand_tracking_predict_gpu --calculator_graph_config_file=mediapipe/graphs/hand_tracking/hand_tracking_desktop_live.pbtxt

  • Run Landmarks_classification/predict.py

About

Real-time Hand Pose Recognition and Dataset Generation With Mediapipe

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published