AI-powered machine vision inspection is crucial for smart factories to ensure efficient production processes and maintain high product quality, as modern factories generate vast amounts of visual data from sensors and cameras deployed throughout the production floor. Transmitting this data to a centralized cloud for AI training and inferencing can strain network bandwidth, lead to delays, and raise concerns about data security.
Solution
Edge AI servers and appliances can process data locally for AI inferencing and training, ensuring data privacy and optimizing bandwidth for smoother data flow and faster response times. Today, we can leverage an edge AI server and appliance for real-time machine vision inspection in diverse applications, such as:
- Errors in the assembly, such as missing parts, misaligned parts, faulty configuration
- Visual defects on surfaces, including scratches, dents, cracks, blowholes, welding defects
- Recognition and verification of colors on any surface, like color irregularities, faulty color configuration
- Recognition, verification, and monitoring of alphanumerical characters, including letters, logos, and data matrix codes
Lanner provides edge AI servers and appliances built to meet the requirements of edge AI inferencing and training for factory production vision inspection.
Edge AI Inference
The Lanner EAI-I730 is a high-performance, modular Edge AI appliance designed to process large amounts of visual data for real-time AI inferencing at the edge. It offers the following features to meet the needs of today's production machine vision:
- AI Acceleration:
The EAI-I730 provides four PCIe slots, including 2x PCIe16 and 2x PCIe4, to support NVIDIA GPU cards and the Falcon H8 PCIE AI accelerator, enabling on-premise, low-latency AI applications.
- Rich I/O:
Equipped with various ports, including 2x 2.5G RJ45, 8x USB 3.2, 2x COM, and 4x DIO, this edge AI appliance is suitable for connecting multiple sensors, cameras, PLCs (Programmable Logic Controllers), RTUs (Remote Terminal Units), and HMIs (Human-Machine Interfaces).
- Storage:
The appliance offers 4x hot-swappable 2.5" HDD/SSD bays with RAID configurations (0, 1, 5, and 10) to accommodate AI frameworks and libraries.
Edge AI Training
The ECA-5540 is a NVIDIA-certified Edge AI server designed to accelerate AI training at the network edge. Key features include:
- High Compute Power:
Powered by the 5th Gen Intel® Xeon® Scalable Processors, the ECA-5540 delivers twice the capacity gains generation over generation within the same power envelope and up to an additional 20% power savings with integrated acceleration.
- AI Acceleration:
The ECA-5540 provides multiple slots for GPU card acceleration, including 1x FHFL PCIex16 slot, plus 2x LP PCIex7 slots or 1x FHHL PCIex8 slot.
Related Articles
- Machine Vision AI Appliances for Factory Product Inspection
- Enabling Automated Packaging & Sorting with Machine Vision Systems