Cattle headcount software used to determine numbers present in political demonstrations


ByDr. Tim Sandle
March 16, 2024

The crowd braved low temperatures in the Belgian capital to the sound of drums and other musical entertainment - Copyright AFP John MACDOUGALL
\
One of the areas of regular dispute relates to estimating the number of people present at a demonstration. Often the figures of march organisers, the police, and the media differ. Can innovative technologies help with this?

Or, perhaps, in the current US political climate, can software challenge the over-sated claims of political figures like Donald Trump when it comes to assessing’ the level of support at rallies?

The answer appears to be ‘yes’ and this involves drawing on an unlikely resource. University of São Paulo has utilized cattle headcount software, such as the type of technology often found on farms, in order to determine the number of people present in political demonstrations: cattle counting software – the type of technology that enables farmers to track down their cattle (albeit that drone technology has added an extra dimension to this process in recent years).

As traditional methods tend to overestimate the number of people present at events, the development highlights the importance of adopting data-driven approaches and advanced technologies for more informed and effective decision-making.

This can become part of the complex interplay between social dynamics, media influence, and individual behaviour.

This approach has led to discussions about the accuracy and reliability of traditional crowd-counting methods and the potential impact of these technologies on political campaigns and conflict situations.

Researcher Pablo Ortellado revealed that the software was trained with a database produced from crowd photos from the University of Xangai in China, showing crowds in various situations.

Based on machine learning, researchers meticulously marked the heads in a quadrant of the images, and the software created a counting and recognition pattern. Before adopting such a methodology, crowd-counting error margins were at 30 percent, but this number dropped to 12 percent after the new method was introduced.

Hence such technology can aid the assessment of mass event analysis. This can have implications in various sectors, from public security to voter behaviour analysis in political campaigns.

The generally accepted definition of a mass-gathering event relates to a group of more than 1000 persons gathered at a specific location for a specific period.

For example, with political campaigns and conflict scenarios, accuracy in crowd counting can have a significant impact, such as understanding a candidate’s true reach and support in political campaigns can influence campaign strategies and political decision-making.