Abstract:
Consistently, a lot of populace accommodates firearm related brutality everywhere. In this work, we foster a PC based completely computerized framework to recognize fundamental deadly implements, especially handguns. Late work in the field of profound learning and move learning has shown huge advancement in the space of item identification and acknowledgment. We have executed YOLOv4 ”You Only Look Once” object location model via preparing it on our custom data set. The preparation results affirm that YOLO V4 beats YOLO V3 and traditional convolutional neural network. Moreover, concentrated GPUs or high calculation assets were not needed in that frame of mind as we utilized transfer learning for preparing our model. Applying this model in our observation framework, we can endeavor to save human existence and achieve decrease in the pace of murder or mass killing. Moreover, our proposed system can likewise be carried out in top of the line observation and security robots to distinguish a weapon or dangerous resources to stay away from any sort of attack or endanger to human existence.
