Towards Urban Public Safety: A Literature Review of Deep Learning Approaches for CCTV-based Violence Detection

Violence detection in video surveillance systems is a critical challenge for ensuring public safety in smart cities.  This study presents a comprehensive analysis of deep learning architectures and transfer learning techniques for violence detection, evaluating their performance across benchmark datasets, including Hockey Fight, Movies, and UCF-Crime.  ResNet50, MobileNetV2, ConvLSTM, DenseNet121, and Xception models are compared in terms of accuracy, computational efficiency, and real-world applicability.  Results highlight ResNet50 as the most effective model, achieving 98.89\% accuracy on UCF-Crime, while MobileNetV2 excels in real-time applications.  Key challenges such as dataset bias, occlusion handling, and ethical considerations are discussed.  The study provides actionable insights for deploying scalable violence detection systems in smart city surveillance networks, balancing accuracy, efficiency, and privacy.

violence detection
глибоке навчання
transfer learning
video surveillance
smart cities
spatiotemporal analysis
  1. Tyagi S., Bagga V.  Deep learning solution for real-time violence detection in video streams.  Advances in AI for Biomedical Instrumentation, Electronics and Computing. CRC Press (2024).
  2. Siddique L. A., Junhai R., Reza T., Khan S. S., Rahman T.  Analysis of Real-Time Hostile Activity Detection from Spatiotemporal Features Using Time Distributed Deep CNNs, RNNs and Attention-Based Mechanisms.  Preprint arXiv:2302.11027 (2023).
  3. Kaur G., Singh S.  A Deep Learning Framework for Violence Detection in Videos Using Transfer Learning.  Advanced Computing, Machine Learning, Robotics and Internet Technologies.  100–111 (2024).
  4. Ghalley A., Abdelsalam A.,Dombola W.  Violence Detection in Automated Surveillance using CNN.  2024 4th International Conference on Intelligent Technologies (CONIT).  1–6 (2024).
  5. Choqueluque-Roman D., Camara-Chavez G.  Weakly Supervised Violence Detection in Surveillance Video.  Sensors.  22 (12), 4502 (2022).
  6. Choudhary R., Solanki A.  Violence Detection in Videos Using Transfer Learning and LSTM.  Advances in Data Computing, Communication and Security.  51–62 (2022).
  7. Ilyas A., Obaid S., Bawany N. Z.  Deep Learning for Violence Detection in Surveillance: The Role of Transfer Learning and Pre-Trained Models.  2023 24th International Arab Conference on Information Technology (ACIT).  1–8 (2023).
  8. Santos F. A. O., Durães D., Marcondes F. S.  Efficient Violence Detection Using Transfer Learning.  Highlights in Practical Applications of Agents, Multi-Agent Systems, and Social Good. The PAAMS Collection.  65–75 (2021).
  9. Vosta S., Yow K. C.  A CNN-RNN Combined Structure for Real-World Violence Detection in Surveillance Cameras.  Applied Sciences.  12 (3), 1021 (2022).
  10. European Road Safety Observatory.  https://road-safety.transport.ec.europa.eu/system/files/2023-02/ff_pedes....
  11. Loce R. P., Bernal E. A., Wu W., Bala R.  Computer vision in roadway transportation systems: a survey.  Journal of Electronic Imaging.  22 (4), 041121 (2013).
  12. Partika F. B. D. M.  Simple Approach for Violence Detection in Real-Time Videos Using Pose Estimation With Azimuthal Displacement and Centroid Distance as Feature.  International Journal of Computer Vision and Image Processing (IJCVIP).  12 (1), 9 (2022).
  13. Temkar R., Bhatia G., Keswani M., Tawade S., Chaudhary T., Karira R.  Automated Violence Detection in Surveillance Networks with Deep Learning.  Advances in Nonlinear Variational Inequalities.  28 (1s), 83–93 (2024).
  14. Suba N., Verma A., Baviskar P. V., Varma S.  Violence detection for surveillance systems using lightweight CNN models.  7th International Conference on Computing in Engineering & Technology (ICCET 2022).  23–29 (2022).
  15. Provath M. A.-M., Deb K., Jo K. H.  XAI-Driven Lightweight Multiscale ConvLSTM Architecture for Video Violence Detection.  2024 IEEE 33rd International Symposium on Industrial Electronics (ISIE).  1–6 (2024).
  16. Thanki K. D., Patel B. V., Malukani S.  A Comprehensive Comparative Analysis of Deep Learning Architectures for Suspicious Activity Detection in Video Surveillance.  International Journal for Research in Applied Science \& Engineering Technology.  12 (IX), 295–305 (2024).
  17. Islam Z., Rukonuzzaman M., Ahmed R., Kabir M. H., Farazi M.  Efficient Two-Stream Network for Violence Detection Using Separable Convolutional LSTM.  2021 International Joint Conference on Neural Networks (IJCNN).  1–8 (2021).
  18. Jain N., Gupta V., Tariq U.  Fast Violence Recognition in Video Surveillance by Integrating Object Detection and Conv-LSTM.  International Journal on Artificial Intelligence Tools.  32 (03), 2340018 (2023).
  19. Chakraborty S., Zahir S., Orchi N. T., Hafiz M. F., Shamsuddoha A. O. M., Dipto S. M.  Violence Detection: A Multi-Model Approach Towards Automated Video Surveillance and Public Safety.  2024 International Conference on Advances in Computing, Communication, Electrical, and Smart Systems (iCACCESS).  1–6 (2024).
  20. Venkatesh B., Satyam Y., Sandhya N.  Study on Deep Learning Techniques for Finding Suspicious Violence Detection in a Video Surveillance.  International Journal on Recent and Innovation Trends in Computing and Communication.  11 (9), 3641–3647 (2023).
  21. Wang X.  Research on Video Surveillance Violence Detection Technology Based on Deep Convolution Network.  2022 International Conference on Information System, Computing and Educational Technology (ICISCET).  347–350 (2022).
  22. Mohammadi H., Nazerfard E.  Video violence recognition and localization using a semi-supervised hard attention model.  Expert Systems with Applications.  212, 118791 (2023).
  23. Srivastava A., Badal T., Singh R.  Real Life Violence Detection in Surveillance Videos using Spatiotemporal Features.  IC3-2021: Proceedings of the 2021 Thirteenth International Conference on Contemporary Computing.  262–266 (2021).
  24. Aremu T., Li Z., Alameeri R., Khan M., El Saddik A.  SSIVD-Net: A Novel Salient Super Image Classification Detection Technique for Weaponized Violence.  Preprint (Version 2), (2023).
  25. Selvaraj J., Anuradha J.  Violence Detection in Video Footages Using I3D ConvNet.  Innovations in Computational Intelligence and Computer Vision.  63–75 (2022).
  26. Sahay K. B., Balachander B., Jagadeesh B., Kumar G. A., Kumar R., Parvathy L. R.  A real time crime scene intelligent video surveillance systems in violence detection framework using deep learning techniques.  Computers and Electrical Engineering.  103, 108319 (2022).
  27. Adithya H. H., Lekhashree H. J., Raghuram S.  Violence Detection in Drone Surveillance Videos.  Smart Trends in Computing and Communications.  703–713 (2023).
  28. Sittalatchoumy R., Pushpa S. E. P., Karthick R., Rafiq Azharudheen M. A., Dhiyanesh K. V.  Transfer Learning based Video Surveillance for Abnormal Activity Detection.  2023 7th International Conference on Intelligent Computing and Control Systems (ICICCS).  419–423 (2023).
  29. Al-Madani A. M., Mahale V., Gaikwad A. T.  Real-Time Detection of Crime and Violence in Video Surveillance using Deep Learning.  Proceedings of the First International Conference on Advances in Computer Vision and Artificial Intelligence Technologies (ACVAIT 2022). 431–441 (2023).
  30. Gadelkarim M., Khodier M., Gomaa W.  Violence Detection and Recognition from Diverse Video Sources.  2022 International Joint Conference on Neural Networks (IJCNN).  1–8 (2022).
  31. Pratama R. A., Yudistira N., Bachtiar F. A.  Violence Recognition on Videos Using Two-stream 3D CNN with Custom Spatiotemporal Crop.  Preprint (Version 1), (2022).
  32. Siddique L. A., Junhai R., Reza T., Khan S. S., Rahman T.  Analysis of Real-Time Hostile Activity Detection from Spatiotemporal Features Using Time Distributed Deep CNNs, RNNs and Attention-Based Mechanisms.  Preprint arXiv:2302.11027 (2023).
  33. Praveen K., Shahina S., Mrudula P., Mohamed H., Anu P.  Video Surveillance System and Violence Detection.  IRO Journal on Sustainable Wireless Systems.  5 (2), 138–149 (2023).
  34. Mohammed H., Elrefaei L.  Detecting Violence in Video Based on Deep Features Fusion Technique.  Preprint arXiv:2204.07443 (2022).
  35. Singh S., Dewangan S., Krishna G., Tyagi V., Reddy S., Medi P. R.  Video Vision Transformers for Violence Detection.  Preprint arXiv:2209.03561 (2022).
  36. Chidananda K., Kumar A.  VidAnomalyNet: An Efficient Anomaly Detection in Public Surveillance Videos Through Deep Learning Architectures.  International Journal of Safety and Security Engineering.  14 (3),  953–966 (2024).
  37. Patel M.  Real-Time Violence Detection Using CNN-LSTM.  Preprint arXiv:2107.07578 (2021).
  38. Rendón-Segador F. J., Álvarez-García J. A., Enríquez F.  ViolenceNet: Dense Multi-Head Self-Attention with Bidirectional Convolutional LSTM for Detecting Violence.  Electronics.  10 (13), 1601 (2021).
  39. Abdullah M.  Violence Detection using Vision Transformer and Explainable AI (2024).