Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{177298,
        author = {SHIVA CHANDAN BODIGE and MS.NEELAM and VARA PRASAD VADLAKONDA and GUTHIKONDA SRINIVAS},
        title = {Detecting and categorizing suspicious/criminal activities from surveillance footage},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {12},
        pages = {438-446},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=177298},
        abstract = {In recent years, increased demands for more heightened public safety activities have pushed surveillance systems forward to obtain real-time detection and reaction of criminal activities. This work focuses on developing a deep learning-based system aimed at the automatic identification of three types of crime video using classifications, namely explosions, shootings, and fights, using Convolutional Neural Networks (CNNs). The traditional video surveillance method relies on human monitoring of many video feeds over a long period of time, which mostly makes the person monitoring it get tired of watching and miss event occurrence. This proposed system exploits the deep learning power to process video data automatically; that is, the occurrence of threats would be identified promptly and correctly. This system integrates several features that are important to usability and security. A user-friendly interface lets the authorized personnel log into the system via a One-Time Password (OTP) in order to access the dashboard. The video footage can be accessed along with alerts on criminal activity, which has been detected. The deep learning model, constructed on the InceptionV3 architecture, is trained on a rich set of video clips where explosions, shootings, and fights occur, thus enabling the model to differentiate between normalcy and suspicious behavior with very high precision. Additionally, there is the provision for an admin dashboard to manage users, update the dataset, and monitor the system, ensuring that the model should have efficacy as new data arises. This project proves the capability of deep learning in achieving the automation of crime detection while assuming the practical needs of security operation by providing a comprehensive, secure solution. Focusing real-time analysis and user accessibility, this system is a huge leap forward in the area of video surveillance, with a scalable deployment tool deployable in environments of all kinds that may increase public safety and improve response times.},
        keywords = {suspicious human activity, criminal activities, Detection, Face recognition, Inceptionv3, CNN, Deep Learning, Image processing, Frames extraction.},
        month = {May},
        }