Abstract: — Based on current circumstances around the world, there is major need of smart surveillance for security to detect weapons and alert the public. The objective here is to develop a prototype which will visually detect the weapons in real time videos and alert upon successful detection. We have worked around this using an algorithm called YOLO-V3. We used a custom dataset of weapons to implement this. The model was implemented to detect weapons in different scenarios with varied parameters like shapes and scales. Further we have integrated the same with twilio api which will send a call alert upon successful detection

Keywords—YOLOV3; Armament detection

| DOI: 10.17148/IARJSET.2021.8732