Parkinson's Disease (PD) is a chronic condition of the nervous system characterized by neuronal degeneration, primarily in the substantia nigra part of the brain. In 2016, a total of 120,000 cases of Parkinson's disease were recognized and documented. Nevertheless, specialists estimate that there are still undetected cases due to external variables such as medical expense and diagnosis accuracy. Because of the slow course of symptoms, early detection and diagnosis of PD have become a challenge in the medical field. Due to the loss of dopamine-producing neurons, patients face reduced motor activities. The symptoms are mainly divided into two categories: i) motor symptoms such as tremors, and slowness of movement; and, ii) non-motor symptoms such as insomnia, sadness, etc. Tremor is one of the first and most prevalent symptoms of Parkinson's disease, and ignoring it at this point could have a serious detrimental influence on the patient's health. In this work, a collection of Deep Learning (DL) models such as DenseNet, Resnet 15, and VGG 16 are used to predict Parkinson's disease by using hand-drawn images and by using DaTscan brain images. The proposed model has achieved maximum accuracy of 91% for hand-drawn images and 95% for DaTscan brain images.