The prevalence of populist discourse in the grassroots movements were largely neglected in quantitative populism studies. By using 9.5 million Facebook posts that were shared in 892 active Facebook groups of Yellow Vests, this research aims to throw light on the prevalence of populist discourse across different political branches of the movement. With the help of supervised text classification method, our aim is to identify whether social media posts shared on public Facebook groups of the Yellow Vest Movement contain the three essential elements of populism: anti-elitism, people-centrism and moralized politics. In compliance with the semantic triplet of populism, we intend to enrich computational methodology by using automatized text classification with supervised machine learning. By detecting social media posts containing the elements of populism, we argue that, in Facebook groups of Yellow Vest Movement in which the members are on the extreme fringes of the left-right spectrum, people tend to use more populist discourse than the moderate ones.