Video vigilance has been implemented at many public places such as streets or buildings to prevent robberies and to obtain proofs from crimes. This information is not always publicly available since has a direct relation with people’s privacy. Another commonly used approach to monitor the security of a city, used by smart cities is the Volunteered Personal Information (VPI) where people provide information through different applications and computational solutions focused in solving urban problems. How to use the above mentioned tools for increase the smart cities security? One solution could be the use of video and volunteered personal information of their citizens. The present approach arises from the people's necessity to have proofs when unfortunately a robbery occurs. In the present approach a deep neural network is implemented in order to identify people and faces inside images that could be analyzed when a robbery occurs, the deep neural network is trained and proved with volunteered personal information (VPI) open source datasets and compared with some other recognition algorithms in order to determine its precision and usability. The deep neural network makes use of bounding boxes to identify people and faces from images. This project is part of a bigger one where the deep neural network would be implemented on a wearable device able to take images in real time when a robbery is happening to its user, hence the need for developing a deep neural network dedicated to the people and faces recognition.