Aayush Kubba, aayushkubba.727@gmail.com
Kamalendu Das, ee.kd13@gmail.com
Mayank Verma, er.mayank.verma1991@gmail.com
Shweta Mayekar, shweta.mayekar695@gmail.com
Souvik Jana, souvikjana1993@gmail.com
Current surveillance and control systems still require human supervision and intervention. Thus far, previous work has mostly focused on weapon-based detection within infrared data for concealed weapons. By contrast, we are particularly interested in the rapid detection and identification of weapons from images and surveillance data. We reformulate this detection problem into the problem of minimizing false positives and solve it by building the key training data-set guided by the results of a deep Convolutional Neural Networks(CNN) classier, then assessing the best classification model under two approaches, the sliding window approach and region proposal approach. Hence we have used Faster RCNN, YOLO and Mask RCNN to build our model. The most promising results are obtained by Mask RCNN based model. We also have validated our model using loss function** to assess the performance of a detection model.
Presentation Link -> https://docs.google.com/presentation/d/1LqsclbcXt5wuGK8mVzsfMmE5cuOsiWnAYZ0flnVxnaE/edit?usp=sharing