Paper Key : IRJ************126
Author: Rithesh K T,Yogaprakash M G,Aruna Bali,Deepu A B,Inchara P
Date Published: 09 Mar 2024
Abstract
In the contemporary world, Video Surveillance holds a crucial role, leveraging advancements in technologies such as artificial intelligence, machine learning, and deep learning. These innovations contribute to the development of sophisticated systems capable of discerning various suspicious behaviors in real-time image monitoring. Given the inherent unpredictability of human behavior, distinguishing between normal and suspicious activities poses a significant challenge. This paper introduces a classification system for human activities, categorizing them into normal (e.g., sitting, walking, jogging, hand waving) and suspicious (e.g., running, boxing, fighting). The classification is achieved through the utilization of convolutional neural networks, extractiing high-level features from images. The convolutional network's classification, along with the final pooling layer result, is considered to make the ultimate prediction, especially in the context of AI-based suspicious activity detection and criminal case identification from video input.Keywords: AI based suspicious activity detection ,Video input , Criminal case detection
DOI LINK : 10.56726/IRJMETS49934 https://www.doi.org/10.56726/IRJMETS49934