Threat Detection and Rescue using Machine Learning

The present work aims to design and implement Digital Image Processing, which is a Numpy and LSTM- based detection kit. The combination has plenty of applications such as hand gesture recognition, sign language translators, home security systems, traffic control technologies, intruder recognition systems, etc. The current investigation utilizes the application for threat detection and rescue system. We have implemented a detection kit for hand gesture tracking and identification of the hand sign which signifies ‘threat’ that consists of two major units namely hardware and software. The hardware part comprises a camera that captures frames in the desired detection range. The detection and identification of the threat is recognized using the following software units such as Anaconda Navigator, Jupyter Notebook, NumPy library, TensorFlow, OpenCV, and MediaPipe. At first, the inputdata is trained for the key-points identification and then the camera captures the frame and compares it with the trained data set. When the captured key point which is hand-signed by the victim is similar to the key points of the trained data set, the threat is detected. The frames are re-evaluated thrice for an accurate result. This is then notified to the officials for rescue. The major advantage of this novel work is the less re-evaluation time and efficient frame process in a very short duration of time. It is also a low-cost system that produces a highly reliable result. This detection will certainly help to save a life and assure a protected life for people in society.