Aim: 50% to 80% of 2 and 4 wheeler owners keep their engines operational while waiting at a red signal. This idling of engines leads to heavy avoidable fuel wastage, vehicular engine damage, respiratory problems, release of greenhouse gases and contributes towards global warming. The present system utilizes microcontrollers to set a fixed traffic signal sequence with present time allotment. In order to release emergency vehicle human assistance is required at the traffic signal controller. We suggest a machine learning based model running on a dual arduino and GPU that continuously monitors and manages the traffic junction using video fed through a CCTV camera. Also running in tandem is a microphone-Arduino setup to monitor for emergency vehicles. These combined allow the elimination of the need for a traffic cop or other human interaction.