Performed crime detection on crime data manually collected from Google images as well as Imagenet database.
Crime detection is one of the highly useful applications in the fields of deep learning, as this helps in curbing the crime and increasing the safety of people. Developed an object detection model using YOLO Darknet to detect harmful weapons such as gun and knife, in the hands of a person. This ensures prevention of false alarms by detecting just the gun or a knife. Also trained the model to detect a person lying down on the road, which helps 911 emergency to reach on time to save life of the person.
Trained the model for 40000 iterations using pretrained weights file convolutional layer : http://pjreddie.com/media/files/darknet19_448.conv.23
