Projects

Using Python Jupyter, we label ECG signals as normal, left bundle branch blockage, right bundle branch blockage, premature ventricle contraction, or atrial premature beat using machine learning and the TensorFlow library. It used the open-source MIT-BIH Arrhythmia Database containing a .csv file for raw ECG signal and a .txt file for annotations.

Database: https://www.kaggle.com/datasets/mondejar/mitbih-database/data

Database information: https://www.physionet.org/content/mitdb/1.0.0/

Previous work has used the CNN model identification but in this code, we will use the LSTM model as it better handles time sequence data as an ECG raw signal would be.

Through Jupyter Notebook we developed a machine-learning CNN model using Pytorch to label the direction of eyes in images. It uses the open-source data from the eye dataset on Kaggle posted by kayvanshah. 

Data set: https://www.kaggle.com/datasets/kayvanshah/eye-dataset

The original intent for the dataset was to automate a wheelchair, but we are repurposing it for future uses in eye tracking.

The data consists of four folders and comprises a total of 14.5k images with the labels close_look: images of a closed eye, left_look: images of the eye looking left, right_look: images of the eye looking right, forward_look: images of the eye looking straight.