A.I People Counter


This is an ongoing project using Tensorflow A.I platform made by google, The aim of this RCNN A.I is to be able to run in live camera footage from a Trade show or Marketing event.

The A.I detects the presence of People and then runs the RCNN network on the detected people as they cross the R.O.I Line (Red line in image), As people pass this counting point the People counter in the top left is incrimentaly increased. These results are fed into a CSV file adding what direction of travel is, Estimated speed. The network is being run on Videosurvalance camera in a high street environment.

The idea for deployment is to have a camera mounted above the exit door‘s at large trade shows or events, as people walk under the camera the A.I counts the attendee’s & then outputs the result.

This data is then fed into a CSV file database allowing the event organizer to build in realtime feedback on the event.

This type of deep data is like gold dust for automated feedback which can then be run into a interactive dashboard giving  graphs for attendee numbers, Thsi Ai is designed to be teamed with out other Gender,Emotion,Age AI to give a complete event classification system for analytics (to create a demographic of end user trends).

The final dashboard then generates graphs and visual statistics (Analytics) allowing to evaluate the event attendee’s.

This can run on edge hardware directly at the event location (High end hardware) or streamed to the cloud from a dumb edge device just streaming video to the AI on-line.



This project is related to our computer vision classification system.

This Project Included

  • Tensorflow AI
  • Python
  • RCNN A.I Deep learning
  • Custom trained A.I Model and Dataset

Here’s a short video clip: