MedementiaX

MedementiaX is a pill-identifying software and pill reminder software. Its machine-learning components were created using TensorFlow Keras. The UI is made using ipythonwidgets.

Training files instructions

For classification.ipynb which is the notebook for the training of our Classifying model for MedementiaX, no file uploads is required

For classification.ipynb which is the notebook for the training details of our Object Classifying model for MedementiaX, no file uploads is required

For PillDetection.ipynb, which is the notebook for the training details of our Object Detection Model for MedementiaX, testing_V2.zip needs to be manually uploaded to the runtime.

Instructions for running the Full Implementation of MedementiaX

1.Upload detector2.h5 and prediction.h5 manually 2.Run all the setup cells

1. Run "PLEASE RUN TO INPUT VALUES FOR MEDICINE QUANTITIES" to input the quantities of medicine to be intaken by the dementia patient 4.Run "FULL IMPLEMENTATION" to run the GUI and upload and check the medicine.

Link to google drive

Google Drive link to code If that doesn't work, try this link: https://drive.google.com/drive/folders/1ZfOsQg_132jW-eOsdEnIvk0BRYgT3bqX?usp=sharing

Google Drive link to slides If that doesn't work, try this link: https://docs.google.com/presentation/d/1KJEdOfNX_vKed23IUvHYwLTBEuiRbALylwbo8nRiW1A/edit?usp=sharing

Challenges we ran into

Labelling images manually is hard, so image detection was limited.

Accomplishments that we're proud of

1. Got a GUI to work in google collab
2. Got image detection and bounding boxes to work decently
3. Object classification works really well

Built With

• python