Technical data:

Videoperformance with Face Detection algorithm, Digital Video, 1920 x 1080p, 2020.

Studies to Deceive a Machine Vision I - Sergio Venancio

Computer Vision is an imperfect instrument of automated judgement on digital images, widely used in contexts of surveillance and biased control. Studies to Deceive a Machine Vision are a series of performatic experiments with Adversarial Attacks, that is, they trick or deceive a Machine Learning trained model by exploring situations which were not considered during its development/training. These experiments, or depictions, were planned from simple body gestures, using head, hands and objects within close reach. The gestures are meant to remember that our bodies are full of identities and complexities that no other vision can have a full grasp of.