* Introduce .
For manufacturing companies, quality control systems have long relied on visual inspection. Traditional machine vision systems may not be able to distinguish defects and defects with high variability among similar parts. Harnessing the power of AI and Deep-learning technology helps solve this problem by providing results with highly accurate detection. Data-driven AI enables automated fault checking with greater flexibility and accuracy, while reducing maintenance costs.
* Challenge .
In this case, the client is a company that develops robotic imaging equipment solutions. It is not flexible enough to update the existing defect testing system to recognize new types of defects. To leverage AI technology for real-time visual defect inspection, the system needs significant computing power at the edge and large storage capacity to store large numbers of images captured from multiple wires. production lines, as well as enough bandwidth to handle data transmission.
* Solution .
The AI system AIR-300 perfectly meets the requirements of customers. After the robotic arm pinpointed the location of the thermos cup and picked it up to take a 360-degree image, the captured images were sent to the AIR-300 for real-time analysis where it could be determined. Identify defective products immediately.
Complex real-time AI inference and high-performance computing have been realized on a local AIR-300 device with power from Intel Xeon / Core i3/i5/i7 CPUs and 1x PCIex16 graphics card support. high performance graphics. In terms of I/O and data storage capacity, the AIR-300 is equipped with 4x GbE 4x RS-232/422/485 ports and supports up to 20TB data capacity with 4x 2.5 SATA III HDDs Provides adequate bandwidth and storage capacity to meet application needs.
The AIR-300 is powered with a built-in 850W power supply, so customers don’t need to worry about providing additional external power. The AIR-300 can also be used as a local training server when the error checking system needs to be updated to test new products. The integrated vision system sends the captured images back to the AIR-300 to retrain the AI model.To change the AI defect testing system from cup inspection to paper bag inspection, customers simply need to prepare training dataset on defect types paper bags have, retry new AI models on AIR -300, then deploy the trained models on the AIR-300. With the ability to retrain AI, updating the error checking system no longer requires the support of professional engineers, which is inherently quite time-consuming and expensive.