International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 11 | Nov 2025
p-ISSN: 2395-0072
www.irjet.net
Artificial Intelligence Based Surveillance Systems: A Survey, Challenges and Future Trends Abdullah Al-Atwi¹, *, Fayez Alkhaibari¹, Khalid Al-Malki1, Yasser Al-Juhani1, Hasan Shtiewi1 1. College of Computing, Fahad Bin Sultan University, Saudi Arabia
*(Corresponding Author) ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract –AI-based surveillance systems leverage computer
vision, machine learning, and IoT sensors to enable real-time monitoring, intelligent analysis, and automated threat response, enhancing security and operational efficiency in smart cities and public spaces. However, their successful implementation hinges on addressing critical cybersecurity and privacy challenges, including secure development frameworks and ethical handling of sensitive data like location information and medical records. This work analyzes these dual imperatives, providing researchers with insights to develop more secure and privacy-preserving surveillance solutions, ultimately fostering trustworthy smart city ecosystems that balance technological advancement with fundamental rights protection.
4. Decision-Making
In terms of components, an AI-based surveillance system consists of five components that are integrated together to achieve the goal, where the input of a given component will be used as output to the next component [2]. Table 1 summarizes the components.
AI-based surveillance systems use computer vision, machine learning, and IoT sensors to automatically monitor, analyze, and respond to activities in real time. These systems enhance security, automate threat detection, and improve operational efficiency in smart cities, businesses, and public spaces [1].
Table -1: Key Components of AI-based Surveillance Systems. Component Cameras & Sensors
1.2 Framework of AI-Based Surveillance Systems A typical AI surveillance system follows a pipeline that consists of five stages [2], as described below: 1. Data Acquisition Sources: CCTV cameras, drones, IoT sensors, facial recognition scanners, license plate readers. Data Types: Video feeds, thermal imaging, audio, motion detection signals.
|
Impact Factor value: 8.315
Edge Devices
Process data locally (low latency)
AI Models
Analyze data for threats Store & analyze bulk data
User Interface
Monitor & control the system
2. 3. 4. 5.
Object Detection (YOLO, Faster R-CNN). Facial Recognition (DeepFace, ArcFace).
© 2025, IRJET
Capture visual/audio data
Cloud/Server
1.
Noise reduction (e.g., stabilizing shaky footage). Frame extraction (converting video into analyzable images).
3. AI Processing & Analysis
Function
Each component uses different technologies enabling it to achieve its function. They are as follows:
2. Preprocessing
Cloud/edge storage for forensic analysis. Indexed databases for fast search (e.g., finding a suspect across multiple cameras).
1.2 Key Components of AI-Based Surveillance Systems
1. INTRODUCTION
Real-time alerts (e.g., gun detection, unauthorized access). Automated responses (e.g., locking doors, notifying authorities).
5. Storage & Retrieval
Key Words: Surveillance, Security, Privacy, Agents, Medical Records.
Behavioral Analysis (anomaly detection using LSTM networks).
|
Cameras & Sensors: IP cameras, LiDAR, thermal imaging. Edge Devices: NVIDIA Jetson, Raspberry Pi + AI accelerators. AI Models: YOLOv8, OpenPose, DeepSORT. Cloud/Server: AWS Rekognition, Azure AI. User Interface: Dashboards (e.g., Milestone XProtect).
ISO 9001:2008 Certified Journal
|
Page 472