Introduction
Artificial Intelligence (AI) is revolutionizing video security, integrating into operational workflows and providing context, precision, and analytical capabilities to functions like perimeter protection, video search, and event detection. AI-powered cameras have become a significant growth driver in the physical security and IoT markets.
The adoption of AI in the industry has grown steadily over the years. Now, large-scale AI models are seen as the next technological leap, with the potential to transform not only security but also operational efficiency and decision-making across multiple industries.
Large-Scale AI in Video Security
Leveraging vast multimodal data and advanced architectures, large-scale AI not only identifies objects or events but also interprets complex relationships, contexts, and patterns. This enables the detection of subtle anomalies and behavioral changes that traditional rule-based or conventional AI approaches fail to perceive, enhancing the strategic value of video security for businesses.
Survey on AI Adoption in Video Security
To understand the current implementation and growth potential of large-scale AI in video security, asmag.com and Hikvision conducted a joint survey analyzing its implementation and future prospects.
The results identify key pain points that organizations aim to resolve with these advanced models and sectors expected to see significant impact. The analysis also reveals challenges that need to be overcome for wider adoption of recent AI technologies.
Key Findings
- Efficiency Operational Improvement: 73.4% of respondents highlighted increased operational efficiency.
- Reduction in Human Error: 62% pointed to a decrease in human error.
- Enhanced User Experience: 55.4% mentioned improved end-user experience.
The areas perceived as most in need of AI-driven innovation are directly linked to automating and simplifying operational tasks prone to errors and time-consuming.
- Video Search and Forensic Research: 56% see potential in video search and forensic research.
- Object and Event Detection: 39.1% anticipate benefits in object and event detection.
- Real-Time Alerts and Response: 34.3% expect improvements in real-time alerts and response capabilities.
AI-driven solutions like Hikvision’s AcuSeek NVRs, based on large-scale AI models, enable video searches using natural language, facilitating immediate event location without extensive manual reviews. Advanced AI-powered cameras like the DeepinView X series distinguish relevant activity from background noise, reducing false alarms by up to 90% compared to conventional AI cameras.
Contextual real-time alerts, driven by large-scale AI, also enable quicker and more informed responses, optimizing decision-making in critical situations. Moreover, respondents identified opportunities for innovation in system configuration and resource optimization (25.9%) and alarm setup and filtering (22.8%), pointing to benefits extending beyond daily operations to security architecture design and continuous improvement.
Sector-Specific Impact
More than half of the respondents (51.2%) identified critical infrastructure as the sector where large-scale AI capabilities will be most relevant, with a significant margin over other sectors.
- Transportation: 40.3%
- Industrial and Manufacturing: 38.5%
The Museo Nacional de Yogyakarta in Indonesia exemplifies this trend, where Hikvision’s AcuSeek system, based on AI, accelerated incident searches and enabled intelligent alerts, reducing response times and enhancing artwork protection.
While retail (21.6%) shows caution due to price sensitivity, the distribution of responses across verticals indicates that large-scale AI will have a broad impact, albeit at varying paces depending on the sector.
Adoption Barriers
The cost of large-scale AI devices remains the primary obstacle to faster adoption, with 60% of respondents identifying it as a barrier. Privacy and regulatory compliance concerns (57%) and integration with legacy systems (53%) follow closely.
Regional analysis reveals distinct differences: 78.9% in Asia cite costs as the main challenge, while privacy and regulatory compliance concerns lead in Europe (81.1%) both within and outside the EU.
Although concerns about model complexity and novelty persist, their impact appears lesser. Only 33.1% of respondents mention a lack of technical knowledge as a barrier, and 18% point to client resistance or skepticism.
As more organizations gain direct experience with these technologies, such hesitations tend to dissipate. Less than a quarter of participants stated their organization has no plans to adopt large-scale AI (9.6%) or maintains a skeptical stance (12.6%). Instead, 38.5% indicated active exploration of its potential, and 37.3% reported already testing or using it, confirming a predominantly favorable attitude toward large-scale AI in video security.
