Flood areas, Person identification, computer vision, COCO Human dataset, Instagram.
Floods become the most common and severe natural disasters in the world due to extreme climate change. In addition to causing serious damage to the economy (human assets) they cause great loss of human life which has resulted in the deaths of people. Early detection is important in providing timely response to prevent damage to property and health. It is therefore important to use all available technologies, including global visibility, prevention and mitigation. On the other hand, focusing on the actions that will be taken immediately after the onset of the flood is very important. Person identification and tracking is a popular and widely used research field in computer vision. There are many security and safety features such as search and rescue, surveillance, driver assistance systems, or automatic driving. Previous flood detection methods use special satellite imagery. In this project, we propose a real-time way of using deep neural network algorithm based on video content feeds analysis from surveillance cameras, the most common and readily available these days. We show that an open CV is a more efficient and faster method of localization, recognition and practice in the COCO Human dataset. When a person is found, the system will take the captured image and post it on social media such as Instagram. This enables not only a team of rescuers who will be busy at the time but also ordinary people close to the area to provide assistance to those people found and rescue them by saving their lives.