Parkinson’s Disease (PD) is an always-incremental untreatable and irremediable neurodegenerative disease. It is a progressive and intensifying nervous system untidiness that mainly transpires at the people, who are in their 60s and becomes substandard with time. It is one of the enfeeble diseases that has no particular medicine or surgery till date. So, it becomes a significant challenge to identify PD at the earliest to prevent harm. PD detection has become an in-demand area of research and study since last decade. Different emerging technologies and approaches have been adapted by researchers in the identification and detection of this disease. In this paper, we have collected the patients’ handwritten drawing dataset from an open source platform. The dataset consists of drawings of both the patients and the healthy people. We have applied the DenseNet, ResNet50, VGG16, MobileNet and Xception model, which are CNN architectures, to achieve the model’s performance with regard to accuracy, sensitivity, F1 score and specificity. This research consequence will be helpful for early PD classification and future research works in the domain of healthcare.