Detection of security flaws and risk situations. We detect security flaws and risk situations, such as the absence of PPE, occupation of escape routes, movement of suspended loads, collision routes and presence in a restricted area. For these reasons, to generate alerts in real time, we process images and videos, using Machine Learning techniques for computer vision. In addition, we also create computer graphics videos of rare and undesirable events, such as accidents, to create synthetic databases used in Machine Learning algorithms.