Intelligent detection systems are crucial for ensuring safety in public places such as schools, sports venues, shopping malls, and crowded hospitals. The global security detection market was valued at $1 billion in 2023 and is expected to exceed $2 billion by 2030. Traditional security measures, such as passive cameras and metal detectors, have limitations, including slow queue speeds and reliance on post-event evidence. The crowded environment and various prohibited items complicate inspections. To address these challenges, customers have adopted AI-driven security check solutions to improve efficiency, reduce waiting times, and enhance user experience.
Project Challenges
Developing AI-driven security checks poses challenges, especially for outdoor events like marathons, which require cost-effective, easy-to-deploy, and seamlessly integrated security systems for on-site safety assurance in a short time. For AI security applications, a hardware system equipped to enhance edge processing and rich multifunctional I/O for connecting devices and sensors is essential. In addition to hardware configuration, developers must pre-train detection models, making the selection of the appropriate model for real-time security checks a critical consideration. Even when pre-trained models are ready, they often need to be converted into a specific format compatible with the target AI processor or GPU. Finally, ensuring the seamless deployment of the converted model onto the AI system is a crucial step.
-
The system provides LAN connectivity for cameras and DIO functionality for anomaly alerts or data collection;
- AI integrates easily into existing systems without additional hardware modifications;
- Supports various pre-trained models for effective hazardous material detection;
Market BackgroundIn this case, the customer utilized Advantech’s AIR-120 Edge AI Inference System, paired with the Hailo-8 powered M.2 AI acceleration module EAI-1200, to achieve real-time security checks. The EAI-1200 offers 26 TOPS of AI processing capability and low power consumption, designed specifically for real-time edge inference tasks. This solution allows for rapid detection of prohibited items such as ammunition, hazardous materials, sharp objects, and weapons.
(Figure: Case Architecture Diagram)The AIR-120 Edge AI Inference System is a fanless box PC based on Intel Atom, featuring LAN ports for connecting two cameras, M.2 expansion for integrating the EAI-1200 AI acceleration module, and digital I/O for data collection. Advantech further enhances this system with its Edge AI SDK, providing inference benchmarking tools, runtime SDK integration, and remote monitoring and management capabilities. Additionally, Hailo’s Model Zoo offers pre-trained models, the Dataflow compiler supports model conversion, and TAPPAS provides pre-trained examples, ensuring seamless deployment and optimization of security check applications.Project Outcomes
- Achieved a 50% reduction in setup costs compared to metal detectors, while increasing detection speed by 10 times.
- Rapid upgrade to AI-supported security checks, enhancing real-time luggage inspection efficiency, processing up to 4,000 people per hour.