Implementation of Realtime design of crowd Enumeration via tracking using AI system

Crowd enumeration can help to evaluate and count the number of visitors to a place. There are many reasons that spana wide range of applications, from security considerations, optimization of operations to efficiency in profitability. In the paper, we propose to develop a prototype for implementing a high frame rate, low processing environment, high performance, and highly efficient real-time crowd enumeration system. The latest method for object detection is deep learning. When it comes to deep learning or machine learning, performance and computation are the key parameters. In our model, there is a provision to schedule the model for the required amount of time. In our work, we are using mobilenet SSD as an object detector to detect humans. It is a preprocessed, highly efficient, and light weight model which can run on low power device like jetson nano and is cost-efficient unlike others. The advantage of our model is if there is overcrowding in a specified location with known capacity, alarm is enabled.