In this paper, we propose an alert system to identify suspects on the scene using mobile devices. This project's objective is to give the police officers a mobile application to receive alerts in real-time; it will help improve the current identification time by using facial recognition. This proposal is due to the increase in Peru's criminal acts that affects 27.3% of the population who has suffered more than one criminal act. Also, the current identification time is between 5 and 10 min. This alert system consists of 1. The collection and training of images of criminals; 2. IP video device configuration; 3. Sending data from the Python service to the application; 4. The process of receiving results in the application. The alert system was validated in a public area of the Rímac district in Lima, Peru. Preliminary results showed that the alert system reduces the current identification time by 91.46% and has an efficiency of 90% based on the tests performed.