A Video Question Answering framework for women’s safety in Metro Surveillance

AI video analytics solutions have already proven effective at airports, bus stations, major thoroughfares, etc. In the context of the metro networks, these solutions are of great utility as they can holistically cover the entire network including metro trains, platforms, concourse areas, entry/exit points and even parking lots. However real-time crowd management in metro surveillance has rarely been addressed. The proposed model puts forward a complete women’s safety framework through Video question-answering technology for Kochi metro surveillance. The proposed model is a deep-learning-based video question-answering model for real-time metro surveillance specifically for women crowd analytics based on age and profession ensuring women’s safety. The model incorporates advanced deep learning techniques for age and gender-based classification of persons especially women commuters along with a dashboard User Interface where visualization of multiple statistics is possible in one go. A Video Question Answering (VideoQA) framework that highlights answering the queries related to metro surveillance data. The proposed dashboard view provides the added advantage of managing multiple crowd-based question answering in one place and aids in quick decision making leading to executing advanced safety measures on time. Our model could achieve superior performance using custom metro surveillance datasets.