Using DataFlow pipelines from pandas to do classification over AlexNet like CNN. Using TENSORFLOW-KERAS and SCI-KIT LEARN on NVIDIA K80 GPU by the University at Albany. Over the NIH CHEST X-RAY Data.set for classification of Diseases.
Primary File: NIH.py
API used: Tensorflow Keras Pandas Scikit-Learn
About the data: The data was obtained from NIH wesbsite. And when Unzipped they all fell into the Images folder in .png format. Some of the data had 3 dimension. Unlike, my course work where I did cv2.imread, I have used Keras to make a continouos flow of Images using the flow_from_dataframe.
Other files: nihtf.py is a pure tensorflow implementation using the DEPRECATED Tensorflow layers API as they are deprecated in TF2.0. The default layer building API for Tensorflow from Tensorflow 2.0 is TENSORFLOW-KERAS. The work also shows the implementation of the Dataset API that is used to quickly make the Data Input Pipeline for the Neural Network.