The demonstration brought together thousands of demonstrators on Paulista Avenue, in Sao Paulo, on Sunday 25th
OR act called by the former president Jair Bolsonaro (PL) this Sunday, 25th, brought together thousands of demonstrators TO Paulista Avenue. The main estimates regarding the audience present differ from each other: the Secretariat of Public Security (SSP) states that there are 600 thousand people were traveling and, if we consider those present in the region adjacent to Paulista, the projection goes to 750 thousand; Already the monitor of digital political debateresearch project at the University of São Paulo (USP), it is estimated that there are around 185 thousand people they were on the avenue at the height of the protest.
The indices guarantee the event one of the largest attendances ever recorded in a political demonstration in Paulista. But sectors of Bolsonarism were not satisfied with the statistics published by both the Secretariat and the USP group. The data was expected to surpass the level reached in Dilma Rousseff’s (PT) pro-impeachment act in 2016, which did not happen.
Federal deputy Eduardo Bolsonaro shared on The publication does not mention the source of the number, and Eduardo Bolsonaro, who did not participate in the protest, does not even mention the origin of the poll.
In another publication in X, the son of the former president directly criticized the USP, whose study group predicted the number of donations at 185 thousand. “The USP knows how to count how many people are on the street just as the little machine counts the votes,” Eduardo Bolsonaro said, referring to electronic ballot boxes.
OR Estadao they queried both sources to understand the methodology of each investigation. Professor Pablo Ortellado, coordinator of the USP group, explained that the tool used processes the data using artificial intelligence. With the help of aerial images, researchers quantify the number of heads (and therefore individuals) present in a given crowd.
The Secretariat also uses aerial images, but the calculation is not carried out by counting people, but by multiplying the observed surface area by an “occupation coefficient”.
-vbcdf57rt1di.jpg)
-rhm5j0wnwz75.jpg)
Culture of “big numbers”
The Political Debate Monitor estimates participation in political protests from August 2022. The USP group’s calculation always differs from projections released by other entities, such as secretariats or police bodies. According to the coordinator of the initiative, the difference lies in the criteria adopted.
“A political culture of grandiose numbers has been created, not only in political demonstrations but on other occasions, such as the Virada Cultural or carnival blocks,” said Pablo Ortellado, professor of Public Policy Management at USP. “These are very large numbers that don’t hold up based on certain criteria,” said the study coordinator who estimated the audience for Sunday’s event.
For Ortellado the difference lies in a more refined criterion and greater precision in the calculation. “We have to get used to methodologically validated rulers who will provide smaller numbers,” he said. The professor also believes that “smaller numbers” do not mean reduced public acts. “Fifty thousand is a huge demonstration; a hundred thousand is a gigantic demonstration.”
Understand the USP method
The USP group’s calculation was done with 43 photographs of the protest. Eight recordings were selected from these images, in order to cover Avenida Paulista in its entirety, without overlapping the areas included in each photo. Subsequently, a method called “Point to Point Network” was used, which consists of counting the number of heads present in each photograph.
Since numerous recordings are made throughout the duration of the protest, it is also possible to calculate the moment of greatest concentration of people during the event, the so-called peak of the demonstration. In the case of Sunday’s protest, the group calculated that the peak occurred at 3 p.m. This estimates the number of people present on each stretch of public road and how many individuals were present at the moment of greatest crowding.
-rhm5j0wnwz75.jpg)
The technology uses artificial intelligence. Despite having the latest data processing technologies, the tool had to be manually “trained” to recognize heads in crowd images. According to Ortellado, the effort also involved a group of USP students to perform the corrections manually.
The first few times the method was implemented by the USP group, the margin of error exceeded 30%. The technology is the same but, after two years of measuring the country’s major political mobilisations, the computer program operates at levels of precision more satisfactory to researchers. Today the margin of error is 12%.
The Secretariat uses the “occupancy coefficient”
The Public Security Secretariat, after being asked to detail the index released on Sunday evening, says that it is based on information that combines “aerial images” with “information from teams on the ground”. In possession of a coefficient indicating the density of people in a given area, the police calculate how many people were in the region. “The technique essentially consists in multiplying the occupied surface by the occupation coefficient”, reads the SSP note.
The calculation is carried out by the Operations Center of the Military Police of the State of São Paulo (Copom) and the software used for the projection is called Copom Online, a tool for internal use of the company. The Secretariat did not inform about the margin of error of the methodology, but said that in all types of public meetings, public estimates are made, “to support operational decisions”.
Source: Terra
Rose James is a Gossipify movie and series reviewer known for her in-depth analysis and unique perspective on the latest releases. With a background in film studies, she provides engaging and informative reviews, and keeps readers up to date with industry trends and emerging talents.
