For many years, the detection of failures in flexible pavement was developed with traditional methods such as: PCI, VIZIR and observation, however, today new technologies such as the use of AI (artificial intelligence) have been developed due to their efficiency and precision. through deep learning (Machine Learning), which is related to image processing, to achieve object detection and its respective analysis. The purpose of the research is to use AI, through the YoloV5 software to fulfill this purpose, taking the Evitamiento Sur road as the study area, in the section between the streets Óvalo Musical (Jr. Atahualpa) and Avenida Industrial, this has a scope Applicative and of an experimental nature, a non-probabilistic sample was shown, based on the two most common faults present in flexible pavement: crocodile and fissure.