NOTE Work has changed to new Repository: https://github.com/djp3/Research---Cerebral-Palsy-Detection.git
Using OpenCV, AWS ML, and AWS Sagemaker we attempt to predict signs of cerebral palsy in premature babies
Install Anaconda For Windows, For Linux, For Mac
-
Open an ancaconda prompt
-
Type
conda create -n yourenvironmentname
-
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
-
Open an anaconda prompt
-
Navigate to your environment that you are using, if it is seperate from the base environment
-
Type
conda install opencv
-
Type
conda install pymysql
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
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
- Done. The python script is in
- 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
- 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.
- 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
- Done. The python script is in
- Ability to convert depth frames in the database into depth-flow images
- Not done. First task for summer students
- 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.