Multimodal Women Safety and Analytics Using OpenCV and OpenVINO

Safety is a basic human right, but the safety of women is still a challenge that the whole world is concerned about, and urgent effective solutions are required. This paper presents an intelligent safety tool based on artificial intelligence, to be deployed in two stages: smart surveillance and user-side applications, including mobile apps and websites. The system will make use of technologies such as the Intel OpenVINO^TM Toolkit and OpenCV for real-time gender detection, crowd density analysis, and gesture recognition. The system can send an SOS through gesture recognition in an emergency, which can enable services like Geocoder and Nominiatim to share the user's location in multiple emergency contacts along with a loud alarm. It also contains a multilingual SOS trigger button within the mobile application and website, community and support of the clients, a resource and information repository. The solution is developed to be deployed using the existing CCTV cameras and smartphones in a cost-effective and scalable way. Regarding benchmark performance, the proposed AI model is almost negligible; detection tasks get completed within milliseconds.