This document presents a study on using vibration sensors and machine learning methods for occupancy detection. It discusses current energy issues in buildings and the need for an occupancy detection system. It describes using vibration sensors as an alternative to other sensor types. The study uses two wireless accelerometers to collect vibration data from a hallway and classroom as people walk by. Features are extracted from the data and a neural network is used to classify the number of occupants. The neural network model achieves over 90% accuracy in detecting 1-6 occupants. The study concludes neural networks provide the best results for occupancy detection compared to other machine learning models.