Increasing Pedestrian Safety is currently an issue that most countries are facing as pedestrian security is of prime concern. This paper aims to propose a technique that utilizes the camera data collected for each street by Montreal’s government. The collected data is used to train a neural network using simulation and TensorFlow, which then identifies pedestrian patterns that are observed during each time of the day. These patterns can then aid in enhancing pedestrian security measures in the chosen streets. The neural network is trained to identify the number of pedestrians in the streets, peak hours at which pedestrian density is maximum, streets which have most pedestrian density throughout the day, and various other patterns described in this work. This data is then used to control the Pedestrian and traffic lights, issue warnings onto the pedestrians’ smartphones in the vicinity of the risk area etc. This task is achieved by utilizing Internet of Things (IoT), Cloud Services, and micro-controllers installed near street lights on streets. Thus, this work not only promotes pedestrian safety but also acts as an aid to developing Smart Cities. This work makes use of the image data freely provided by the city of Montreal. It also discusses the advantages, weaknesses, limitations, and future scope of the above-mentioned proposed technique.