DrIDENT: THE APP FOR PILL RECOGNITION

A model designed to recognize solid pharmaceuticals

THE DATASETS

Dataset 1:

• https://datadiscovery.nlm.nih.gov/Drugs-and-Chemicals/Computational-Photography-Project-for-Pill-Identif/5jdf-gdqh/about_data (provided by the National Library of Medicine) Dataset 2:
• https://www.kaggle.com/datasets/trumedicines/pharmaceutical-tablets-dataset (provided by Gaurav Dutta on Kaggle)

NOTE: Due to the size of the dataset the datasets must be downloaded if the code should be run.

How to run the code:

• Repository File Overview:
  • DrIDENT-code.py

    needs the following files to run:

    • metadata: table.csv from Dataset 1, Training_set.csv from Dataset 2
    • images in folders: original from Dataset 1, train from Dataset 2 After downloading the datasets please insert the path to both the metadata and the images in order to run the code.

    Please make sure the following packages are installed:

    • numpy
    • pandas
    • cv2
    • os
    • math
    • matplotlib.pyplot
    • tensorflow
    • pickle
    • scikit-learn
    • streamlit
    • tf2onnx
    • PIL
    • io
  • DrIDENT.pptx
  • DrIDENT.pdf