Paper Key : IRJ************946
Author: Ankit Singh,Adhyatm Mishra ,Purva Patil,Sahil More,Anita Mhatre
Date Published: 02 Apr 2024
Abstract
Rise in criminal activities in suburban and metropolitan areas gives a threat to social stability and people's safety. It becomes inevitable to put an end to these practices. In previous times it was a burdensome to monitor these activities. Now with help of Machine Learning and AI, identifying and categorizing suspicious human behavior becomes easier with the help of a CNNLSTM hybrid LRCN model. Conversion of image sequences into labels, probabilities, descriptions as well as learning of visual attributes from video frames etc. becomes much easier with the help of this technique. Suspicious activities like fighting, robbery and firing are identified by the system. The proposed system uses Keras in Python, and Tensorflow based on the LRCN model. It gathers video frames and compares them with the trained models for monitoring the problems with an accuracy of approximate 91.5%. Additional features of this system includes the use of AI to raise alarm, send email as an alert to the concerned authorities and create report of detected suspicious activities in a simplified PDF format, all prompting to take necessary actions.
DOI Requested