djp3 / Research---Cerebral-Palsy-Detection

Using OpenCV, AWS ML, and AWS Sagemaker we attempt to predict signs of cerebral palsy in premature babies

NOTE Work has changed to new Repository: https://github.com/djp3/Research---Cerebral-Palsy-Detection.git

Research---Cerebral-Palsy-Detection

Using OpenCV, AWS ML, and AWS Sagemaker we attempt to predict signs of cerebral palsy in premature babies

To Run the Dense_Optical_Flow.py file:

Install Anaconda For Windows, For Linux, For Mac

Create a virtual environment:

1. Open an ancaconda prompt

2. Type conda create -n yourenvironmentname

3. Activate your new env with source activate yourenvironmentname

Creating Virtual Environments with Anaconda

Note: You can also create a new environment using the GUI provided in the Anaconda Navigator

Note: You will also need to install packages such as OpenCV and pymysql to compile the Dense_Optical_Flow program

Install required packages:

1. Open an anaconda prompt

2. Navigate to your environment that you are using, if it is seperate from the base environment

3. Type conda install opencv

4. Type conda install pymysql

Final Steps

Install spyder using the Anaconda Navigator. I used Spyder2.3.8 but the latest version should be fine

Use spyder to open, run, or edit the Dense_Optical_Flow.py program

Turnover stuff

Things to give Prof. Patterson

• Database access:
  • Done. Database is on AWS under Patterson's root credentials
• Accelerometer
  • Ability to convert raw accelerometer in the database to generated in the db Python script that runs outside of AWS to capture the raw accelerometer data and put the generated data into AWS.
    • Done. The python script is in 2019_Summer_Workflow/01_Accelerometer_Raw_To_Generated/feature_extraction.ipynb
  • Ability to train an XGBoost model on accelerometer generated data and test it
    • Done. The jupyter notebook is in SageMaker called nested_kfold_xgb_from_db.ipynb: https://xl.notebook.us-east-1.sagemaker.aws/notebooks/nested_kfold_xgb_from_db.ipynb
• Video
  • Ability to convert raw rgb frames in the database to optical flow images in the db. A Python script that runs outside of AWS to capture the raw accelerometer data and put the generated data into AWS.
    • Done. The python script is in 2019_Summer_Workflow/02_Video_Raw_To_Generated/dense_optical_flow.py
  • Ability to convert depth frames in the database into depth-flow images
    • Not done. First task for summer students