Cardiovascular disease is the predominant cause of morbidity and mortality globally. The Myocardial Infarction is one of the disorders that most significantly affects Indian patients. Digital cardiopathy plays a crucial role in the identification of cardiovascular disease through the use of artificial intelligence. The most prevalent non-invasive instrument is the ECG, which is widely utilized for the detection of cardiovascular diseases globally. In middle and lower-income nations, the majority of available ECG reports are in paper format, which can be readily converted into digital form using smartphone cameras or other photographic devices. This study presents a ResNet50-based deep learning framework for the detection of four types of cardiovascular diseases from 12-lead ECG images: Normal Sinus Rhythm (N), Abnormal Heartbeat (AH), Myocardial Infarction (MI), and History of Myocardial Infarction (HMI). This study also presents a novel 12-lead ECG imaging dataset originating from India. The proposed deep learning architecture is being trained and verified using the aforementioned dataset, combined with the only accessible foreign ECG dataset from University of Management and Technology, Pakistan. The suggested framework achieves an accuracy of 98.23% and a precision of 98.31%, above the existing state of the art. We subsequently deployed the proposed model on the Jetson Nano Edge device, incorporating a GUI for near real-time inference. The implemented model demonstrated an accuracy of 98%, with a near real-time inference duration of 3.4 ms.