Fire-Safe is a drone mountable device equipped with a thermal camera and a human detection feature. The thermal camera provides a live feed of thermal images to the user and is resilient to heat and smoke. The human detection feature is done by using a YOLO model pretrained on the COCO dataset and using transfer learning we use the weights for our machine learning model that detects humans and surrounds them via a bounding box.
In the scene of a house-fire, time is of the essence. Firefighters often spend a large amount of time preparing to enter a house fire and then searching for signs of life at house-fire sites. Fire-Safe is a drone mountable device that has the ability to detect human life via a machine learning algorithm.
The goal of Fire-Safe is to use the time it takes for firefighters to prepare to search and find human life, expediting the process of saving human life in house fire settings.
Provide a live feed of thermal images, Have the device be wireless
Have the ability to detect a human via a bounding, Be light enough to be mounted onto a drone
Team 13's FPR Slides