This project involved the implementation of computer vision on filtered images for state prediction. Image classification performed on segmented images. The aim was to understand the scene and environment surrounding the car.

A new network based on modern CNN models was developed which could improveaccuracy and reduce the false-positive rate.

In this project consolidation of differentobservations was performed by using Using Bayesian Networks and HMM. To make a decision under uncertainty, a general form of probability was used which is called Evidence theory.

To further improve the temporal denoising, a Hidden Markov Model was implemented Implementing Hidden Markov Model for Evidence Theory was also implemented.

The result was comparable to other machine learning approaches.

In this project medical conditions were predicted and recommended using a web server. The most updated data/measures were used for postural stability predictions. These data were used to run inferences on trained models. The predictive modelswere utilized in Docker Containers in a web server that runs over Flask API.

A Kubernetes platform also employed for server management.