Incept_LSTM : Accession for human activity concession in automatic surveillance. (17th November 2022)

• Record Type: Journal Article
• Authors: Girdhar, Palak; Johri, Prashant; Virmani, Deepali
• Is Part Of: Journal of discrete mathematical sciences & cryptography. Volume 25:Number 8(2022)
• Page Start: 2259
• Page End: 2273
• Publication Date: 2022-11-17
• Subjects: 68U10; Face recognition -- Human activity recognition -- Deep learning -- Behavioral analysis; Computer science -- Mathematics -- Periodicals; Cryptography -- Periodicals; Computer science -- Mathematics; Cryptography; Periodicals; 004.0151
• DOI: 10.1080/09720529.2020.1804132
• Languages: English
• ISSNs: 0972-0529

Abstract

Abstract: Automatic monitoring is increasingly being used today, as it helps to detect various unforeseen events that may cause threat. To provide basic classification data, the surveillance operation is monitored via sensors and cameras. Any unfair situation / person / event is a complex activity to detect. It includes various external as well as internal factors such as: context environment, gestures of the person – movement of the body and the hand, muscle strain, personal identity and demography. Automatic testing uses certain variables for identification in literature .In this paper, Incept_LSTM used for video surveillance using deep learning method called Inception-based LSTM for Human Activity Recognition (HAR). The proposed system aims at improving the system performance by monitoring human activities closely. The approach proposed is using Inception v3 as a deep learning model for extraction of the features. And other unit, LSTM is used to capture the temporal or time series data. The training and evaluation data was empirically tested with an accuracy of 91 per cent on UCF crime dataset.