In India, more than 150,000 people are killed each year in traffic accidents. Which in 2016 logged about 40,000. There is one death every 4 minutes due to a road accident in India. One serious road accident in the country occurs every minute and 16 die on Indian roads every hour. The first hour of such an emergency, called the “golden hour” is widely regarded as the most precious time for saving lives. The 201 Law of Commission had stated, “Doctors point out that at least 50% of fatalities can be averted if victims are admitted to a hospital within the first one hour”. To reduce these types of casualties, many projects have been published. The existing works which are published in various journals can prevent accidents, but in few journals propose methods of reporting accidents to the nearby emergency unit using automated systems having sensors. We are trying to alter the existing system from scratch with a state-of-the-art Accident detection system using Artificial Intelligence. In our proposed system, high resolution traffic cameras placed at specific angles detect the accidents with the help of a special Algorithm. Classification of data is achieved using machine learning algorithms to find the severity of the accident, crossing a predefined threshold of probability compared with a pre-trained database. After finding the severity, it transmits the alert notification to nearby emergency units such as ambulance, firefighters and police. Through this project, we try to overcome the problem using the concept of Deep Learning, and Artificial Intelligence.