search

Unity-Technologies / Robotics-Object-Pose-Estimation

A complete end-to-end demonstration in which we collect training data in Unity and use that data to train a deep neural network to predict the pose of a cube. This model is then deployed in a simulated robotic pick-and-place task.
Apache License 2.0
293 stars 75 forks source link

Resume Training #48

Closed agru97 closed 2 years ago

agru97 commented 2 years ago

How can I resume training when it got stopped during the process?

at669 commented 2 years ago

Initializing a PoseEstimationEstimator will try to load the estimators from file if a checkpoint_file exists: https://github.com/Unity-Technologies/Robotics-Object-Pose-Estimation/blob/7558d74b21f6e22544138c5b5f8525f24ff60c98/Model/pose_estimation/pose_estimation_estimator.py#L69-L72 which calls load(): https://github.com/Unity-Technologies/Robotics-Object-Pose-Estimation/blob/7558d74b21f6e22544138c5b5f8525f24ff60c98/Model/pose_estimation/pose_estimation_estimator.py#L103-L123

Or, if you're using Docker, you may investigate using checkpoints.

Hope this helps!