Uso do reconhecimento facial para busca de fraudes em redes sociais.

Abstract

Faced with the current digital age where almost everything is being systematized, one can find criminals working in this area as well. There has been an increasing number of thefts of social media accounts in exchange for money to return or sell a product that does not belong to the person, fraud with false profiles impersonating other people or establishments and somehow causing harm to those who are innocent. Based on these factors, this work proposes the creation of an algorithm capable of collecting photos through hashtags in instagram, performing facial recognition and classifying if any of these photographs contain the individual who is interested in knowing if there are possible photos that could be causing you some kind of damage or not. In the course of this work, it is observed that to carry out the implementation, internet data collectors (Web Crawler and Web Scraping) are being used together with detection and facial recognition. The algorithms that helped the development are the CNN (Convolutional Neural Network) and the HOG (Oriented Gradient Histogram) for the detection of faces, in addition to making a comparison between them, together with the KNN (K-nearest neighbors) algorithm. ) for their facial recognition.