Unsupervised multilabel image segmentation (color/gray/multichannel) based on the Potts model (aka piecewise constant Mumford-Shah model)

Implements an entire machine learning pipeline to train and evaluate a Random Forest Classifier on labeled gait data for walking. Data generated during the experiment has led to helpful insights in to the problem domain.

This is an Android application that solves Dead reckoning with help of step counter

Smoothing splines for signals with discontinuities

Step Detection using SVM on NURVV trackers

Tool for detecting abrupt step changes in noisy signals