Skip to content

Latest commit

 

History

History
47 lines (24 loc) · 1.91 KB

README.md

File metadata and controls

47 lines (24 loc) · 1.91 KB

Headpose Estimator Apache Mxnet

Head Pose estimator using Apache MXNet. HeadPose_ResNet50_Tutorial.ipynb helps you to walk through an end-to-end work flow of developing a CNN model from the scratch including data augmentation, fine-tuning, saving check-point model artifacts, validation and inference.

Preprocessing head-pose data

Please run the following command first to prepare the input data file.

python2 preprocessingDataset_py2.py --num-data-aug 15 --aspect-ratio 1

HeadPose_ResNet50_Tutorial

Jupyter notebook to develop Headpose Estimator CNN model using Apache MXNet.

HeadPose_ResNet50_Tutorial_Gluon

Jupyter notebook to develop Headpose Estimator CNN model using Gluon.

HeadPose_SageMaker_PythonSDK

Two sets of SageMaker notebooks and entry point scripts to develop the Headpose Estimator model on Amazon SageMaker.

  • HeadPose_SageMaker_PySDK.ipynb: SageMaker notebook to invoke an entry point python script.

  • EntryPt-headpose.py: An entry point python script to train Headpose Estimator model. This entry point script is analogous to HeadPose_ResNet50_Tutorial.ipynb.

  • EntryPt-headpose-wo-cv2.py: The entry point script without cv2.

  • HeadPose_SageMaker_PySDK-Gluon.ipynb: SageMaker notebook to invoke an entry point python script.

  • EntryPt-headpose-Gluon.py: An entry point python script to train Headpose Estimator model. This entry point script is analogous to HeadPose_ResNet50_Tutorial_Gluon.ipynb.

  • EntryPt-headpose-Gluon-wo-cv2.py: The entry point script without cv2.

  • tensorflow_resnet_headpose_for_deeplens.ipynb: SageMaker notebook to invoke an TensorFlow entry point python script.

  • resnet_headpose.py: The TensorFlow main entry point script used for training and hosting

  • resnet_model_headpose.py: TensorFlow ResNet model

testIMs

Sample head images for inference test.

License

This library is licensed under the Apache 2.0 License.