An electrocardiogram (ECG) is an important basis for diagnosing cardiovascular disease, which has a significant impact on heart health. This study suggests combining a number of features to create an effective system for classifying the normal beat (N), premature ventricular contraction (V), supraventricular ectopic beat (S), fusion beat (F), and unknown beat (Q). In order to improve accuracy while minimizing execution time and resource consumption, we have suggested a pretrained deep learning model along with a basic CNN model that fixes classification issues through data preprocessing. We have also evaluated the performance of DenseNet 169 and ResNet 50. Moreover, we also proposed a hybrid deep learning model with a data augmentation methodology. To prove the efficacy of the proposed hybrid model, we have also evaluated a large dataset and achieved an accuracy of 98.91%. The proposed hybrid model performs better in evaluating the discriminative power of heart disease and heartbeat type.