Counting People in Images

Analyse images and automatically count the amount of people / persons which can be recognized therein.
Counting People in Images

Description

Monitoring the total amount of people / persons is an important part of many business processes in industries such as as retail, venue management, or security. CCTV camera systems often already monitor the people flow, but these systems often remain "dumb". A Computer vision AI service can help to analyze CCTV images, detect persons therein and calculate an overall people count.

Problem

A business process requires to measure the total amount of people or vehicles present at a certain location at a certain time. This can involve crowd monitoring, security management or general purpose foot traffic analysis.

Solution

A computer vision AI service is able to detect people entities in images so that a downstream software application is then able to count these detected entities. The reference solution architecture for this use case looks like this:

  1. Take an image from the location where people should be counted (e.g. CCTV)
  2. Preprocess the image(s), i.e. adjust the image size to what is required by the Computer Vision AI service
  3. The preprocessed image will be sent to a Computer Vision AI service of your choice.
  4. In a post-processing step the result will be filtered for person entities and a given confidence level. 
  5. The output can then be summarized as a simple number count or a picture where the detected people labels are highlighted. 

Benefit

Automatic crowd monitoring removes the tedious manual task of counting people in a venue or area and assures automatic data flow with correct time stamps and a selected confidence level for people detection. This process can be automated and adjusted as needed: More pictures per minute / seconds to adjust accuracy and/or higher image quality. 

Video-Tutorial

Solution Overview