Cardiovascular Diseases are of the leading causes of death worldwide. For people recovering from different heart conditions, Cardiac Rehabilitation (CR) is an essential part of the treatment process. During CR, exercise intensity is vital just as the monitoring of patients. Monitoring patients can be facilitated using Computer Vision and Deep Learning. We propose an approach of using Optical Flow and Deep Learning in order to achieve this task. In our approach, we use Optical Flow to track the breathing behavior of patients by tracking the movement of the chest before and after exercising. The resulting tracks are drawn onto a black frame which is a representative of the breathing pattern of the patient. The resulting breathing patterns are then inputted into our ResNet-50 model to be classified. Moreover, we manually gathered our dataset with the aid of a CR expert. We carried out 2 experiments. Firstly, points on the chest of the participants were manually identified using Social LEAP Estimates Animal Poses (SLEAP), these points were then tracked using Optical Flow, and the resulting breathing patterns were inputted into our model resulting in a testing accuracy of 84.38%, a precision of 85.23%, and a recall of 81.52%. Secondly, corner detection was performed using the Features from Accelerated Segment Test (FAST) corner detector. These corners were then tracked using Optical Flow. The resulting breathing patterns of this experiment were inputted into the model which resulted in a testing accuracy of 97.4%, a precision of 98.85%, and a recall of 95.56%.